Verus verification

Extend the Azure Policy parser to handle complete policyRule and
policyDefinition JSON structures, not just standalone constraints.

Policy rule parser (policy_rule.rs):
- Parse top-level { "if": ..., "then": ... } objects
- Extract effect kind (deny, audit, append, modify, etc.) into typed AST
- Parse "details" structurally when it is an object to pull out
  existenceCondition as a first-class Constraint; fall back to opaque
  JSON for non-object details (e.g. append array form)
- Detect duplicate/missing keys for "if", "then", "effect", "details"

Policy definition parser (policy_definition.rs):
- Handle both wrapped ARM envelope ({ "properties": { ... } }) and
  unwrapped (properties-level keys at top level) forms
- Type-extract displayName, description, mode, metadata, parameters,
  and policyRule; everything else goes into extra
- Parse parameter definitions with type, defaultValue, allowedValues,
  and metadata; detect duplicate parameter names
- Duplicate key detection throughout

Grammar documentation (docs/azure-policy/azurepolicy.ebnf):
- Add formal EBNF grammar covering policy-rule, then-block,
  constraints, conditions, all 19 operators, count expressions,
  JSON values, and ARM template expressions

Test harness changes:
- Add parse_level field to YAML test cases: "constraint" (default),
  "policy_rule", or "policy_definition"
- Un-skip three parse_errors cases that needed policy_rule-level parsing
- Add policy_rule.yaml with 12 cases covering all 9 effect kinds,
  existenceCondition, parameterized effects, complex conditions, and
  extra key handling
- Add policy_definition.yaml with wrapped, unwrapped, parameterized,
  missing-policyRule, and duplicate-key error cases

.cargo/config.toml

@@ -0,0 +1,2 @@
+[alias]
+xtask = "run --package xtask --"

.devcontainer/devcontainer.json

@@ -4,11 +4,20 @@
 	"name": "Rust",
 	// Or use a Dockerfile or Docker Compose file. More info: https://containers.dev/guide/dockerfile
 	"image": "mcr.microsoft.com/devcontainers/rust:1-1-bullseye",
+	"customizations": {
+		"vscode": {
+			"extensions": [
+				"ms-dotnettools.csharp",
+				"ms-dotnettools.csdevkit"
+			]
+		}
+	},
 	"features": {
-		"ghcr.io/devcontainers/features/dotnet:2": {},
+		"ghcr.io/devcontainers/features/dotnet:2": {
+			"version": "8.0"
+		},
 		"ghcr.io/devcontainers/features/python:1": {}
 	}
-
 	// Use 'mounts' to make the cargo cache persistent in a Docker Volume.
 	// "mounts": [
 	// 	{

.github/actions/toolchains/rust/action.yml

@@ -0,0 +1,29 @@
+name: rust-toolchain
+description: Setup Rust toolchain with specified version and components
+inputs:
+  toolchain:
+    description: 'Rust toolchain version'
+    required: false
+    default: '1.92.0'
+  components:
+    description: 'Additional components to install'
+    required: false
+    default: 'clippy rustfmt'
+  targets:
+    description: 'Target architectures to install'
+    required: false
+    default: ''
+runs:
+  using: composite
+  steps:
+    - shell: bash
+      run: |
+        rustup override set ${{ inputs.toolchain }}
+        if [ -n "${{ inputs.components }}" ]; then
+          rustup component add ${{ inputs.components }}
+        fi
+        if [ -n "${{ inputs.targets }}" ]; then
+          rustup target add ${{ inputs.targets }}
+        fi
+        cargo --version
+        rustc --version

.github/dependabot.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 # To get started with Dependabot version updates, you'll need to specify which
 # package ecosystems to update and where the package manifests are located.
 # Please see the documentation for all configuration options:
@@ -5,7 +7,95 @@
 
 version: 2
 updates:
+  # All Rust/Cargo directories are grouped into a single entry so that
+  # when a dependency is updated, Dependabot bumps it across the root
+  # workspace AND every binding, preventing version skew.
   - package-ecosystem: "cargo"
-    directory: "/" # Location of package manifests
+    directories:
+      - "/"
+      - "/bindings/ffi"
+      - "/bindings/java"
+      - "/bindings/python"
+      - "/bindings/ruby"
+      - "/bindings/wasm"
     schedule:
       interval: "weekly"
+    commit-message:
+      prefix: "build(deps)"
+    groups:
+      # Bundle all Cargo dependency updates into a single PR. Without this,
+      # dependabot creates a separate PR per directory for the same dependency,
+      # and each individual PR fails to build due to version skew.
+      rust-dependencies:
+        patterns:
+          - "*"
+    # Ignore vendored mimalloc crates; updates are managed manually.
+    ignore:
+      - dependency-name: "regorus-mimalloc"
+      - dependency-name: "regorus-mimalloc-sys"
+
+  - package-ecosystem: "gomod"
+    directory: "/bindings/go"
+    schedule:
+      interval: "weekly"
+    commit-message:
+      prefix: "build(deps)"
+    groups:
+      per-dependency:
+        patterns:
+          - "*"
+
+  - package-ecosystem: "maven"
+    directory: "/bindings/java"
+    schedule:
+      interval: "weekly"
+    commit-message:
+      prefix: "build(deps)"
+    groups:
+      per-dependency:
+        patterns:
+          - "*"
+
+  - package-ecosystem: "nuget"
+    directory: "/bindings/csharp"
+    schedule:
+      interval: "weekly"
+    commit-message:
+      prefix: "build(deps)"
+    groups:
+      per-dependency:
+        patterns:
+          - "*"
+
+  - package-ecosystem: "pip"
+    directory: "/bindings/python"
+    schedule:
+      interval: "weekly"
+    commit-message:
+      prefix: "build(deps)"
+    groups:
+      per-dependency:
+        patterns:
+          - "*"
+
+  - package-ecosystem: "bundler"
+    directory: "/bindings/ruby"
+    schedule:
+      interval: "weekly"
+    commit-message:
+      prefix: "build(deps)"
+    groups:
+      per-dependency:
+        patterns:
+          - "*"
+
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"
+    commit-message:
+      prefix: "ci(deps)"
+    groups:
+      github-actions:
+        patterns:
+          - "*"

.github/workflows/codeql.yml

@@ -0,0 +1,193 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
+name: "CodeQL Security Analysis"
+
+on:
+  schedule:
+    # Run weekly on Wednesdays at 3:17 AM UTC
+    - cron: '17 3 * * 3'
+  workflow_dispatch:
+    # Allow manual triggering
+  push:
+    branches: [ "main" ]
+  pull_request:
+    branches: [ "main" ]
+
+jobs:
+  analyze:
+    name: Analyze (${{ matrix.language }})
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+    permissions:
+      # required for all workflows
+      security-events: write
+      # required to fetch internal or private CodeQL packs
+      packages: read
+      # only required for workflows in private repositories
+      actions: read
+      contents: read
+
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          # Rust analysis for main crate and Rust-based bindings
+          - language: rust
+            build-mode: none
+            working-directory: .
+          # C/C++ analysis for FFI bindings
+          - language: c-cpp
+            build-mode: manual
+            working-directory: bindings/ffi
+          # Python analysis for Python bindings
+          - language: python
+            build-mode: none
+            working-directory: bindings/python
+          # Java analysis for Java bindings
+          - language: java-kotlin
+            build-mode: manual
+            working-directory: bindings/java
+          # Go analysis for Go bindings
+          - language: go
+            build-mode: manual
+            working-directory: bindings/go
+          # C# analysis for C# bindings
+          - language: csharp
+            build-mode: manual
+            working-directory: bindings/csharp
+          # JavaScript analysis for WASM bindings
+          - language: javascript-typescript
+            build-mode: none
+            working-directory: bindings/wasm
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+    # Setup language-specific dependencies BEFORE CodeQL init for proper tracing setup
+    - name: Setup Rust
+      uses: ./.github/actions/toolchains/rust
+
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus
+
+    - name: Fetch workspace dependencies
+      run: cargo fetch --locked
+
+    - name: Fetch FFI crate dependencies
+      if: matrix.language == 'c-cpp' || matrix.language == 'go' || matrix.language == 'csharp'
+      run: cargo fetch --locked --manifest-path bindings/ffi/Cargo.toml
+
+    - name: Fetch Java crate dependencies
+      if: matrix.language == 'java-kotlin'
+      run: cargo fetch --locked --manifest-path bindings/java/Cargo.toml
+
+    - name: Setup Python
+      if: matrix.language == 'python'
+      uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+      with:
+        python-version: '3.10'
+
+    - name: Setup Java
+      if: matrix.language == 'java-kotlin'
+      uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
+      with:
+        distribution: 'corretto'
+        java-version: '8'
+
+    - name: Setup Go
+      if: matrix.language == 'go'
+      uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
+      with:
+        go-version: '1.21'
+
+    - name: Setup .NET
+      if: matrix.language == 'csharp'
+      uses: actions/setup-dotnet@c2fa09f4bde5ebb9d1777cf28262a3eb3db3ced7 # v5.2.0
+      with:
+        global-json-file: ./bindings/csharp/global.json
+
+    - name: Invoke dotnet directly
+      if: matrix.language == 'csharp'
+      run: dotnet --info
+
+    - name: Setup Node.js
+      if: matrix.language == 'javascript-typescript'
+      uses: actions/setup-node@53b83947a5a98c8d113130e565377fae1a50d02f # v6.3.0
+      with:
+        node-version: '18'
+
+    - name: Initialize CodeQL
+      uses: github/codeql-action/init@c10b8064de6f491fea524254123dbe5e09572f13 # v4.35.1
+      with:
+        languages: ${{ matrix.language }}
+        build-mode: ${{ matrix.build-mode }}
+
+    # Install additional build dependencies
+    - name: Install system dependencies
+      if: matrix.language == 'rust' || matrix.language == 'c-cpp'
+      run: |
+        sudo apt-get update
+        sudo apt-get install -y build-essential cmake
+
+    - name: Install Python build dependencies
+      if: matrix.language == 'python'
+      working-directory: ${{ matrix.working-directory }}
+      run: |
+        python -m pip install --upgrade pip
+        pip install maturin[patchelf] pytest
+
+    - name: Setup Ruby
+      if: matrix.language == 'rust' && contains(matrix.working-directory, 'ruby')
+      uses: ruby/setup-ruby@3ff19f5e2baf30647122352b96108b1fbe250c64 # v1.299.0
+      with:
+        ruby-version: '3.4.2'
+        bundler-cache: true
+        working-directory: bindings/ruby
+
+    - name: Install WASM build dependencies
+      if: matrix.language == 'javascript-typescript'
+      run: |
+        cargo install wasm-pack
+
+    # Manual build steps for different languages
+    - name: Build C/C++ bindings via xtask
+      if: matrix.language == 'c-cpp'
+      run: |
+        cargo xtask test-c --release --frozen
+        cargo xtask test-cpp --release --frozen --skip-ffi
+        cargo xtask test-c-no-std --release --frozen --skip-ffi
+
+    - name: Build Java bindings via xtask
+      if: matrix.language == 'java-kotlin'
+      run: cargo xtask test-java --release --frozen
+
+    - name: Build Go bindings via xtask
+      if: matrix.language == 'go'
+      run: cargo xtask test-go --release --frozen
+
+    - name: Build C# bindings manually
+      if: matrix.language == 'csharp'
+      working-directory: ${{ matrix.working-directory }}
+      run: |
+        # Temporary workaround: CodeQL's tracer replaces dotnet with a missing shim when cargo xtask test-csharp runs,
+        # so invoke dotnet directly here until the upstream fix lands.
+        # Ideal command once fixed: cargo xtask test-csharp --release
+        # Build the FFI library that C# bindings access via P/Invoke
+        cd ../ffi
+        cargo build --release --locked
+        cd ../csharp
+        # Restore NuGet packages and build .NET assemblies in release mode
+        dotnet restore Regorus/Regorus.csproj
+        dotnet build Regorus/Regorus.csproj --no-restore /p:Configuration=Release /p:IgnoreMissingArtifacts=true
+
+    - name: Build WASM bindings via xtask
+      if: matrix.language == 'javascript-typescript'
+      run: cargo xtask build-wasm --release
+
+    - name: Perform CodeQL Analysis
+      uses: github/codeql-action/analyze@c10b8064de6f491fea524254123dbe5e09572f13 # v4.35.1
+      with:
+        category: "/language:${{matrix.language}}"

.github/workflows/dependabot-refresh-cargo-lockfiles.yml

@@ -0,0 +1,129 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
+name: dependabot/refresh-cargo-lockfiles
+
+on:
+  pull_request_target:
+    types: [opened, synchronize, reopened]
+    branches: ["main"]
+
+concurrency:
+  group: dependabot-refresh-cargo-lockfiles-${{ github.event.pull_request.number }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  refresh-cargo-lockfiles:
+    permissions:
+      contents: write
+    if: >-
+      github.event.pull_request.user.login == 'dependabot[bot]' &&
+      github.event.pull_request.head.repo.full_name == github.repository
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd
+        with:
+          repository: ${{ github.event.pull_request.head.repo.full_name }}
+          ref: ${{ github.event.pull_request.head.ref }}
+          fetch-depth: 0
+          persist-credentials: false
+
+      - name: Setup Rust toolchain
+        run: |
+          rustup toolchain install 1.92.0 --profile minimal
+          rustup override set 1.92.0
+          cargo --version
+          rustc --version
+
+      - name: Refresh affected Cargo lockfiles
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          base_sha="${{ github.event.pull_request.base.sha }}"
+          head_sha="${{ github.event.pull_request.head.sha }}"
+
+          mapfile -t changed_files < <(git diff --name-only "$base_sha" "$head_sha" -- ':(glob)**/Cargo.toml' ':(glob)**/Cargo.lock')
+
+          if [ "${#changed_files[@]}" -eq 0 ]; then
+            echo "No Cargo manifest or lockfile changes detected."
+            exit 0
+          fi
+
+          declare -A manifests=()
+          for path in "${changed_files[@]}"; do
+            case "$path" in
+              bindings/ffi/*)
+                manifests["bindings/ffi/Cargo.toml"]=1
+                ;;
+              bindings/java/*)
+                manifests["bindings/java/Cargo.toml"]=1
+                ;;
+              bindings/python/*)
+                manifests["bindings/python/Cargo.toml"]=1
+                ;;
+              bindings/ruby/*)
+                manifests["bindings/ruby/Cargo.toml"]=1
+                ;;
+              bindings/wasm/*)
+                manifests["bindings/wasm/Cargo.toml"]=1
+                ;;
+              *)
+                manifests["Cargo.toml"]=1
+                ;;
+            esac
+          done
+
+          for manifest in "${!manifests[@]}"; do
+            echo "Refreshing lockfile for $manifest"
+            cargo metadata \
+              --config 'build.rustc="rustc"' \
+              --config 'build.rustc-wrapper=""' \
+              --config 'build.rustc-workspace-wrapper=""' \
+              --format-version 1 \
+              --all-features \
+              --manifest-path "$manifest" > /dev/null
+          done
+
+          if [[ -n "${manifests[Cargo.toml]+x}" ]]; then
+            echo "Refreshing lockfile for tests/ensure_no_std/Cargo.toml (thumbv7m-none-eabi)"
+            cargo metadata \
+              --config 'build.rustc="rustc"' \
+              --config 'build.rustc-wrapper=""' \
+              --config 'build.rustc-workspace-wrapper=""' \
+              --format-version 1 \
+              --manifest-path tests/ensure_no_std/Cargo.toml \
+              --filter-platform thumbv7m-none-eabi > /dev/null
+          fi
+
+      - name: Commit lockfile refresh
+        shell: bash
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          set -euo pipefail
+
+          mapfile -t lockfiles < <(git ls-files -m -o --exclude-standard -- ':(glob)**/Cargo.lock')
+
+          for lockfile in "${lockfiles[@]}"; do
+            git add "$lockfile"
+          done
+
+          if git diff --cached --quiet; then
+            echo "No Cargo lockfile changes required."
+            exit 0
+          fi
+
+          auth_header=$(printf 'x-access-token:%s' "$GH_TOKEN" | base64 | tr -d '\n')
+          trap 'git config --unset-all http.https://github.com/.extraheader' EXIT
+          git config http.https://github.com/.extraheader "AUTHORIZATION: basic ${auth_header}"
+          git config user.name "github-actions[bot]"
+          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
+          git commit -m "build(deps): refresh Cargo lockfiles"
+          git push origin HEAD:${{ github.event.pull_request.head.ref }}

.github/workflows/dependency-audit.yml

@@ -0,0 +1,66 @@
+name: Dependency Audits
+
+on:
+  pull_request:
+  push:
+    branches: ["main"]
+  schedule:
+    - cron: "0 6 * * 1"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  cargo-audit:
+    name: Cargo Audit (${{ matrix.lockfile }})
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        lockfile:
+          - Cargo.lock
+          - bindings/ffi/Cargo.lock
+          - bindings/java/Cargo.lock
+          - bindings/python/Cargo.lock
+          - bindings/ruby/Cargo.lock
+          - bindings/wasm/Cargo.lock
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+
+      - name: Run cargo audit
+        uses: rustsec/audit-check@v2
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          lockfile: ${{ matrix.lockfile }}
+
+  cargo-deny:
+    name: Cargo Deny (${{ matrix.manifest }})
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        manifest:
+          - Cargo.toml
+          - bindings/ffi/Cargo.toml
+          - bindings/java/Cargo.toml
+          - bindings/python/Cargo.toml
+          - bindings/ruby/Cargo.toml
+          - bindings/ruby/ext/regorusrb/Cargo.toml
+          - bindings/wasm/Cargo.toml
+          - tests/ensure_no_std/Cargo.toml
+          - xtask/Cargo.toml
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+
+      - name: Setup Rust
+        uses: ./.github/actions/toolchains/rust
+
+      - name: Run cargo deny
+        uses: EmbarkStudios/cargo-deny-action@v2
+        with:
+          command: check
+          command-arguments: advisories bans
+          manifest-path: ${{ matrix.manifest }}

.github/workflows/feature-matrix.yml

@@ -0,0 +1,82 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
+# Thorough weekly test of non-default feature combinations.
+# Catches regressions from dependency updates and feature-gating issues
+# that the fast PR CI checks (cargo check only) would miss at runtime.
+name: tests/feature-matrix
+
+on:
+  workflow_dispatch:
+  schedule:
+    # Run at 3:42 AM UTC every Saturday.
+    - cron: "42 3 * * 6"
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  feature-matrix:
+    name: ${{ matrix.name }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          # Bare minimum: validates that the core interpreter works
+          # without any builtins or optional subsystems.
+          - name: minimal (std + arc)
+            features: std,arc
+
+          # Common library usage pattern (issue #595): consumer enables
+          # std + arc + rvm and relies on indexmap/std propagation.
+          - name: library (std + arc + rvm)
+            features: std,arc,rvm
+
+          # New default after removing mimalloc from full-opa.
+          # Ensures all builtins compile without the allocator.
+          - name: full-opa (no mimalloc)
+            features: std,arc,full-opa
+
+          # Binding-style usage: full-opa with the vendored allocator.
+          # Mirrors how ffi/java/python/ruby bindings are built.
+          - name: full-opa + allocator
+            features: std,arc,full-opa,allocator-memory-limits
+
+          # Selective builtins without full-opa: validates that popular
+          # features can be cherry-picked independently.
+          - name: cherry-picked builtins
+            features: std,arc,rvm,regex,time,semver,cache
+
+          # Observability features only: coverage + cache without the
+          # heavier builtins (regex, time, etc.).
+          - name: observability
+            features: std,arc,rvm,coverage,cache
+
+          # Azure Policy adds jsonschema + dashmap; test it compiles
+          # and runs on top of full-opa.
+          - name: azure-policy
+            features: std,arc,full-opa,azure_policy
+
+          # Azure RBAC adds regex + time + net on top of full-opa.
+          - name: azure-rbac
+            features: std,arc,full-opa,azure-rbac
+
+          # no_std with the OPA-compatible feature set: exercises the
+          # spin_no_std codepath and absence of std-only dependencies.
+          - name: no_std
+            features: arc,opa-no-std
+
+    steps:
+    - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+    - name: Setup Rust toolchain
+      uses: ./.github/actions/toolchains/rust
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus-features
+    - name: Fetch dependencies
+      run: cargo fetch --locked
+    - name: Build
+      run: cargo build --no-default-features --features "${{ matrix.features }}" --frozen
+    - name: Test
+      run: cargo test --no-default-features --features "${{ matrix.features }}" --frozen

.github/workflows/miri.yml

@@ -0,0 +1,29 @@
+name: miri
+
+on:
+  workflow_dispatch:
+  schedule:
+    # Run at 6:30 AM UTC every Wednesday
+    - cron: "30 6 * * 3"
+
+jobs:
+  miri-test:
+    name: miri (nightly)
+    runs-on: ubuntu-latest
+    env:
+      MIRIFLAGS: "-Zmiri-disable-isolation"
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+      - uses: ./.github/actions/toolchains/rust
+        with:
+          toolchain: nightly
+          components: miri rust-src
+      - name: Set up Miri
+        run: cargo miri setup
+      - name: Run Miri tests
+        run: cargo miri test -p regorus
+      - name: Run Miri ACI tests
+        run: cargo miri test -p regorus --test aci
+      - name: Run Miri kata tests
+        run: cargo miri test -p regorus --test kata

.github/workflows/pr-extensions.yml

@@ -1,10 +1,15 @@
-name: tests/release
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
+name: tests/release-extensions
 
 on:
   push:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 env:
   CARGO_TERM_COLOR: always
@@ -15,19 +20,18 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v4
-    - name: Build only std
-      run: cargo build -r --example regorus --no-default-features --features "std,rego-extensions"
-    - name: Doc Tests
-      run: cargo test -r --doc --features rego-extensions
-    - name: Run tests
-      run: cargo test -r --features rego-extensions
-    - name: Run example
-      run: cargo run --example regorus --features rego-extensions -- eval -d examples/server/allowed_server.rego -i examples/server/input.json data.example
-    - name: Run tests (ACI)
-      run: cargo test -r --test aci --features rego-extensions
-    - name: Run tests (KATA)
-      run: cargo test -r --test kata --features rego-extensions
-    - name: Run tests (OPA Conformance)
+    - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+    - name: Setup Rust toolchain
+      uses: ./.github/actions/toolchains/rust
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus
+    - name: Fetch dependencies
+      run: cargo fetch --locked
+    - name: Run rego extensions CI suite
       run: >-
-        cargo test -r --test opa --features opa-testutil,serde_json/arbitrary_precision,rego-extensions  -- $(tr '\n' ' ' < tests/opa.passing)
+        cargo xtask ci-release --frozen --features rego-extensions
+        --skip-all-features-build --skip-no-default-features-tests
+        --skip-azure-policy --skip-azure-rbac
+        --opa-features "opa-testutil,serde_json/arbitrary_precision,rego-extensions"

.github/workflows/pr.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: tests/release
 
 on:
@@ -5,6 +7,9 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 env:
   CARGO_TERM_COLOR: always
@@ -15,27 +20,14 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v4
-    - name: Format Check
-      run: cargo fmt --check
-    - name: Build (all features)
-      run: cargo build -r --all-features
-    - name: Build
-      run: cargo build -r
-    - name: Test no_std
-      run: cargo test -r --no-default-features
-    - name: Build only std
-      run: cargo build -r --example regorus --no-default-features --features "std"
-    - name: Doc Tests
-      run: cargo test -r --doc
-    - name: Run tests
-      run: cargo test -r
-    - name: Run example
-      run: cargo run --example regorus -- eval -d examples/server/allowed_server.rego -i examples/server/input.json data.example
-    - name: Run tests (ACI)
-      run: cargo test -r --test aci
-    - name: Run tests (KATA)
-      run: cargo test -r --test kata
-    - name: Run tests (OPA Conformance)
-      run: >-
-        cargo test -r --test opa --features opa-testutil,serde_json/arbitrary_precision  -- $(tr '\n' ' ' < tests/opa.passing)
+    - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+    - name: Setup Rust toolchain
+      uses: ./.github/actions/toolchains/rust
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus
+    - name: Fetch dependencies
+      run: cargo fetch --locked
+    - name: Run release CI suite
+      run: cargo xtask ci-release --frozen

.github/workflows/publish-java.yml

@@ -1,6 +1,9 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: publish-java
 
-on: workflow_dispatch
+on:
+  workflow_dispatch:
 
 permissions:
   contents: read
@@ -32,26 +35,28 @@ jobs:
             os: windows-latest
             extension: dll
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
-      - uses: actions/setup-java@v4
+      - uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
         with:
           java-version: 8
           distribution: "corretto"
-      - uses: dtolnay/rust-toolchain@stable
+      - uses: ./.github/actions/toolchains/rust
         with:
           targets: ${{ matrix.target }}
       - if: ${{ matrix.build_cmd == 'zigbuild' }}
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
         with:
           python-version: "3.11"
       - if: ${{ matrix.build_cmd == 'zigbuild' }}
         run: pip install cargo-zigbuild
-      - run: cargo ${{ matrix.build_cmd || 'build' }} --release --target ${{ matrix.target }}${{ matrix.glibc && format('.{0}', matrix.glibc) || '' }} --manifest-path ./bindings/java/Cargo.toml
+      - run: cargo fetch --locked
+      - run: cargo fetch --locked --manifest-path bindings/java/Cargo.toml
+      - run: cargo ${{ matrix.build_cmd || 'build' }} --release --frozen --target ${{ matrix.target }}${{ matrix.glibc && format('.{0}', matrix.glibc) || '' }} --manifest-path ./bindings/java/Cargo.toml
       - run: mkdir -p native/${{ matrix.target }}
       - run: mv target/${{ matrix.target }}/release/*.${{ matrix.extension }} ./native/${{ matrix.target }}/
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
         with:
           name: native-libraries-${{ matrix.target }}
           path: native/
@@ -61,24 +66,24 @@ jobs:
     runs-on: ubuntu-latest
     needs: build
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
-      - uses: actions/setup-java@v4
+      - uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
         with:
           java-version: 8
           distribution: "corretto"
           server-id: ossrh
           server-username: MAVEN_USERNAME
           server-password: MAVEN_PASSWORD
-      - uses: actions/download-artifact@v4
+      - uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
         with:
           pattern: native-libraries-*
           merge-multiple: true
           path: ./bindings/java/native/
       - run: mvn package
         working-directory: ./bindings/java
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
         with:
           name: built-jars
           path: ./bindings/java/target/regorus-java-*.jar

.github/workflows/publish-python.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 # This file is autogenerated by maturin v1.4.0
 # To update, run
 #
@@ -18,21 +20,30 @@ jobs:
       matrix:
         target: [x86_64, x86, aarch64, armv7, s390x, ppc64le]
     steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
         with:
           python-version: '3.10'
+      - uses: ./.github/actions/toolchains/rust
+
+      - name: Build Python extension
+        run: |
+          cargo fetch --locked
+          cargo clippy --all-targets --no-deps -- -Dwarnings
+          cargo build --release --target ${{ matrix.target }} --frozen
+        working-directory: bindings/python
+
       - name: Build wheels
-        uses: PyO3/maturin-action@63b75c597b83e247fbf4fb7719801cc4220ae9f3 # v1.43.0
+        uses: PyO3/maturin-action@04ac600d27cdf7a9a280dadf7147097c42b757ad # v1.43.0
         with:
           target: ${{ matrix.target }}
-          args: --release --out dist --find-interpreter --manifest-path bindings/python/Cargo.toml
+          args: --release --out dist --manifest-path bindings/python/Cargo.toml --offline --strip
           sccache: 'true'
           manylinux: auto
       - name: Upload wheels
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
         with:
-          name: wheels
+          name: wheels-linux-${{ matrix.target }}
           path: dist
 
   windows:
@@ -41,21 +52,30 @@ jobs:
       matrix:
         target: [x64, x86]
     steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
         with:
           python-version: '3.10'
           architecture: ${{ matrix.target }}
+      - uses: ./.github/actions/toolchains/rust
+
+      - name: Build Python extension
+        run: |
+          cargo fetch --locked
+          cargo clippy --all-targets --no-deps -- -Dwarnings
+          cargo build --release --target ${{ matrix.host.target }} --frozen
+        working-directory: bindings/python
+
       - name: Build wheels
-        uses: PyO3/maturin-action@63b75c597b83e247fbf4fb7719801cc4220ae9f3 # v1.43.0
+        uses: PyO3/maturin-action@04ac600d27cdf7a9a280dadf7147097c42b757ad # v1.43.0
         with:
           target: ${{ matrix.target }}
-          args: --release --out dist --find-interpreter --manifest-path bindings/python/Cargo.toml
+          args: --release --out dist --manifest-path bindings/python/Cargo.toml --frozen --strip
           sccache: 'true'
       - name: Upload wheels
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
         with:
-          name: wheels
+          name: wheels-windows-${{ matrix.target }}
           path: dist
 
   macos:
@@ -64,35 +84,29 @@ jobs:
       matrix:
         target: [x86_64, aarch64, universal2-apple-darwin]
     steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
         with:
           python-version: '3.10'
+      - uses: ./.github/actions/toolchains/rust
+
+      - name: Build Python extension
+        run: |
+          cargo fetch --locked
+          cargo clippy --all-targets --no-deps -- -Dwarnings
+          cargo build --release --target ${{ matrix.host.target }} --frozen
+        working-directory: bindings/python
+
       - name: Build wheels
-        uses: PyO3/maturin-action@63b75c597b83e247fbf4fb7719801cc4220ae9f3 # v1.43.0
+        uses: PyO3/maturin-action@04ac600d27cdf7a9a280dadf7147097c42b757ad # v1.43.0
         with:
           target: ${{ matrix.target }}
-          args: --release --out dist --find-interpreter --manifest-path bindings/python/Cargo.toml
+          args: --release --out dist --manifest-path bindings/python/Cargo.toml --offline --strip
           sccache: 'true'
       - name: Upload wheels
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
         with:
-          name: wheels
-          path: dist
-
-  sdist:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - name: Build sdist
-        uses: PyO3/maturin-action@63b75c597b83e247fbf4fb7719801cc4220ae9f3 # v1.43.0
-        with:
-          command: sdist
-          args: --out dist --manifest-path bindings/python/Cargo.toml
-      - name: Upload sdist
-        uses: actions/upload-artifact@v3
-        with:
-          name: wheels
+          name: wheels-macos-${{ matrix.host.target }}
           path: dist
 
   release:
@@ -100,13 +114,15 @@ jobs:
     runs-on: ubuntu-latest
     # Commented out for initial release.
     # if: "startsWith(github.ref, 'refs/tags/')"
-    needs: [linux, windows, macos, sdist]
+    needs: [linux, windows, macos]
     steps:
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
         with:
-          name: wheels
+          pattern: wheels-*
+          merge-multiple: true
+          path: wheels
       - name: Publish to PyPI
-        uses: PyO3/maturin-action@63b75c597b83e247fbf4fb7719801cc4220ae9f3 # v1.43.0
+        uses: PyO3/maturin-action@04ac600d27cdf7a9a280dadf7147097c42b757ad # v1.43.0
         env:
           MATURIN_PYPI_TOKEN: ${{ secrets.PYPI_API_TOKEN }}
         with:

.github/workflows/publish-wasm.yml

@@ -1,10 +1,13 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: publish-wasm
 
 permissions:
   pull-requests: write
   contents: write
 
-on: workflow_dispatch
+on:
+  workflow_dispatch:
 
 jobs:
   publish-wasm:
@@ -12,11 +15,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
       # Setup .npmrc file to publish to npm
-      - uses: actions/setup-node@v4
+      - uses: actions/setup-node@53b83947a5a98c8d113130e565377fae1a50d02f # v6.3.0
         with:
           node-version: '20.x'
           registry-url: 'https://registry.npmjs.org'

.github/workflows/release-plz.yml

@@ -1,24 +1,29 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: Release-plz
 
 permissions:
   pull-requests: write
   contents: write
 
-on: workflow_dispatch
+on:
+  workflow_dispatch:
 
 jobs:
   release-plz:
     name: Release-plz
     runs-on: ubuntu-latest
+    permissions:
+      contents: write
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
       - name: Install Rust toolchain
-        uses: dtolnay/rust-toolchain@stable
+        uses: ./.github/actions/toolchains/rust
       - name: Run release-plz
-        uses: MarcoIeni/release-plz-action@98b2b45b090aadf18cb662caaf3de6222d98822a #v0.5.60
+        uses: MarcoIeni/release-plz-action@1528104d2ca23787631a1c1f022abb64b34c1e11 # v0.5.128
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           CARGO_REGISTRY_TOKEN: ${{ secrets.CARGO_REGISTRY_TOKEN }}

.github/workflows/rust-clippy.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 # This workflow uses actions that are not certified by GitHub.
 # They are provided by a third-party and are governed by
 # separate terms of service, privacy policy, and support
@@ -16,6 +18,9 @@ on:
     # The branches below must be a subset of the branches above
     branches: [ "main" ]
   workflow_dispatch:
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 jobs:
   rust-clippy-analyze:
@@ -27,28 +32,27 @@ jobs:
       actions: read # only required for a private repository by github/codeql-action/upload-sarif to get the Action run status
     steps:
       - name: Checkout code
-        uses: actions/checkout@v2
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
 
-      - name: Install Rust toolchain
-        uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af #@v1
+      - name: Setup Rust toolchain
+        uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
         with:
-          profile: minimal
-          toolchain: stable
-          components: clippy
-          override: true
+          shared-key: ${{ runner.os }}-regorus
 
       - name: Install required cargo
         run: cargo install clippy-sarif sarif-fmt
 
+      - name: Fetch
+        run: cargo fetch --locked
+
       - name: Run rust-clippy
-        run:
-          cargo clippy
-          --all-features
-          --message-format=json | clippy-sarif | tee rust-clippy-results.sarif | sarif-fmt
-        continue-on-error: true
+        run: cargo xtask clippy --sarif rust-clippy-results.sarif
 
       - name: Upload analysis results to GitHub
-        uses: github/codeql-action/upload-sarif@v1
+        if: ${{ hashFiles('rust-clippy-results.sarif') != '' }}
+        uses: github/codeql-action/upload-sarif@c10b8064de6f491fea524254123dbe5e09572f13 # v3.29.11
         with:
           sarif_file: rust-clippy-results.sarif
           wait-for-processing: true

.github/workflows/test-c-cpp.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/c-cpp
 
 on:
@@ -5,44 +7,38 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 jobs:
   test:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
 
+      - uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch FFI crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/ffi/Cargo.toml
+
       - name: Setup gcc, g++, cmake, ninja
         run: sudo apt update && sudo apt install -y gcc g++ cmake ninja-build
 
-      - name: Workaround to ensure that regorus.h is generated
-        run: |
-          cargo build -r
-        working-directory: ./bindings/ffi
-        
-      - name: Test c binding
-        run: |
-          mkdir bindings/c/build
-          cd bindings/c/build
-          cmake -G Ninja ..
-          ninja
-          ./regorus_test
+      - name: Test C binding via xtask
+        run: cargo xtask test-c --release --frozen
 
-      - name: Test c-nostd binding
-        run: |
-          mkdir bindings/c-nostd/build
-          cd bindings/c-nostd/build
-          cmake -G Ninja ..
-          ninja
-          ./regorus_test
+      - name: Test C (no-std) binding via xtask
+        run: cargo xtask test-c-nostd --release --frozen --skip-ffi
 
-      - name: Test cpp binding
-        run: |
-          mkdir bindings/cpp/build
-          cd bindings/cpp/build
-          cmake -G Ninja ..
-          ninja
-          ./regorus_test
+      - name: Test C++ binding via xtask
+        run: cargo xtask test-cpp --release --frozen --skip-ffi

.github/workflows/test-csharp.yml

@@ -1,26 +1,166 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/csharp
 
 on:
+  workflow_dispatch:
   push:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
-jobs:
-  test:
-    runs-on: ubuntu-latest
+env:
+  VersionSuffix: ${{ github.event_name == 'workflow_dispatch' && 'manualtrigger' || null }}
+
+jobs:  
+  build-ffi:
+    name: 'Build Regorus FFI: (${{ matrix.runtime.target }})'
+    runs-on: ${{ matrix.runtime.os }}
+    strategy:
+      # let us get failures from other jobs even if one fails
+      fail-fast: false
+      matrix:
+        runtime:
+          - os: windows-latest
+            target: x86_64-pc-windows-msvc
+            libpath: |
+              **/release/regorus_ffi.dll
+              **/release/regorus_ffi.pdb
+          - os: ubuntu-latest
+            target: x86_64-unknown-linux-gnu
+            libpath: |
+              **/release/libregorus_ffi.so
+          - os: macos-latest
+            target: aarch64-apple-darwin
+            libpath: |
+              **/release/libregorus_ffi.dylib
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          fetch-depth: 0
+      - uses: ./.github/actions/toolchains/rust
+        with:
+          targets: ${{ matrix.runtime.target }}
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch FFI crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/ffi/Cargo.toml --target ${{ matrix.runtime.target }}
+
+      - name: Build Regorus FFI via xtask
+        run: cargo xtask build-ffi --release --target ${{ matrix.runtime.target }}
+        
+      - name: Upload regorus ffi shared library
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
+        with:
+          name: regorus-ffi-artifacts-${{ matrix.runtime.target }}
+          # Note: The full path of each artifact relative to . is preserved.
+          path: ${{ matrix.runtime.libpath }}
+          if-no-files-found: error
+          retention-days: 1
+
+  build-csharp:
+    name: 'Build Regorus nuget'
+    runs-on: ubuntu-latest
+    needs: build-ffi
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          fetch-depth: 0  
+      - uses: ./.github/actions/toolchains/rust
+      - uses: actions/setup-dotnet@c2fa09f4bde5ebb9d1777cf28262a3eb3db3ced7 # v5.2.0
+        with:
+          global-json-file: ./bindings/csharp/global.json
+
+      - run: echo '${{ steps.stepid.outputs.dotnet-version }}'
+
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Download regorus ffi shared libraries
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
+        with:
+          pattern: regorus-ffi-artifacts-*
+          merge-multiple: true
+          path: ./bindings/csharp/Regorus/tmp
+
+      - name: Display regorus ffi artifacts
+        run: ls -R ./bindings/csharp/Regorus/tmp
+
+      - name: Build Regorus nuget via xtask
+        run: cargo xtask build-csharp --release --clean --artifacts-dir ./bindings/csharp/Regorus/tmp/bindings/ffi/target --enforce-artifacts --repository-commit ${{ github.sha }} --include-symbols
+
+      - name: Upload Regorus nuget
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
+        with:
+          name: regorus-nuget
+          path: |
+            bindings/csharp/Regorus/bin/Release/Regorus*.nupkg
+            bindings/csharp/Regorus/bin/Release/Regorus*.snupkg
+          if-no-files-found: error
+          retention-days: 1
+
+  test-nuget:
+    name: 'Test Regorus Nuget: (${{ matrix.runtime.target }})'
+    needs: build-csharp
+    runs-on: ${{ matrix.runtime.os }}
+    strategy:
+      # let us get failures from other jobs even if one fails
+      fail-fast: false
+      matrix:
+        runtime:
+          - os: windows-latest
+            target: x86_64-pc-windows-msvc
+          - os: ubuntu-latest
+            target: x86_64-unknown-linux-gnu
+          - os: macos-latest
+            target: aarch64-apple-darwin
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
 
-      - uses: actions/setup-dotnet@v4
+      - uses: ./.github/actions/toolchains/rust
 
-      - name: Build
-        run: dotnet build
-        working-directory: ./bindings/csharp/net8.0
+      - uses: actions/setup-dotnet@c2fa09f4bde5ebb9d1777cf28262a3eb3db3ced7 # v5.2.0
+        with:
+          global-json-file: ./bindings/csharp/global.json
 
-      - name: Run
-        run: LD_LIBRARY_PATH=. dotnet run
-        working-directory: ./bindings/csharp/net8.0
+      - run: echo '${{ steps.stepid.outputs.dotnet-version }}'
+
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Download regorus nuget
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
+        with:
+          name: regorus-nuget
+          path: ./bindings/csharp/Regorus/bin/Release
+
+      - name: Display regorus nuget
+        run: ls -R ./bindings/csharp/Regorus/bin/Release
+
+      - name: Run C# tests via xtask
+        run: cargo xtask test-csharp --release --clean --nuget-dir bindings/csharp/Regorus/bin/Release
+
+      

.github/workflows/test-csharp40.yml

@@ -1,28 +0,0 @@
-name: bindings/csharp40
-
-on:
-  push:
-    branches: [ "main" ]
-  pull_request:
-    branches: [ "main" ]
-
-jobs:
-  test:
-    runs-on: windows-latest
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-
-      - uses: actions/setup-dotnet@v1
-        with:
-            dotnet-version: "5.0.x"
-
-      - name: Build
-        run: dotnet build
-        working-directory: ./bindings/csharp/net40
-
-      - name: Run
-        run: dotnet run
-        working-directory: ./bindings/csharp/net40

.github/workflows/test-ffi.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/ffi
 
 on:
@@ -5,18 +7,28 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 jobs:
   test:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
+      - uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch FFI crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/ffi/Cargo.toml
 
       - name: Test FFI
-        run: |
-          cargo build -r
-          cargo clippy --all-targets --no-deps -- -Dwarnings
-        working-directory: ./bindings/ffi
+        run: cargo xtask test-ffi --release --frozen

.github/workflows/test-go.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/go
 
 on:
@@ -5,29 +7,32 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 jobs:
   test:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
-   
-          
-      - uses: actions/setup-go@v5
+      - uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch FFI crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/ffi/Cargo.toml
+
+      - uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
-          python-version: ${{ env.PYTHON_VERSION }}
           architecture: x64
 
-      - name: Build ffi
-        run: cargo build -r
-        working-directory: ./bindings/ffi
-
-      - name: Test go
-        run: |
-          go mod tidy
-          go build
-          LD_LIBRARY_PATH=../../target/release ./regorus_test
-        working-directory: ./bindings/go
+      - name: Test Go binding via xtask
+        run: cargo xtask test-go --release --frozen

.github/workflows/test-java.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/java
 
 on:
@@ -5,33 +7,33 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 jobs:
   test:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
 
-      - uses: actions/setup-java@v4
+      - uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
         with:
           java-version: 8
           distribution: "corretto"
-      - uses: dtolnay/rust-toolchain@stable
+      - uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
 
-      - name: Building binding
-        run: |
-          cargo clippy --all-targets --no-deps -- -Dwarnings
-          cargo build --release --manifest-path bindings/java/Cargo.toml
+      - name: Fetch Java crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/java/Cargo.toml
 
-      - name: Build jar
-        run: mvn package
-        working-directory: ./bindings/java
-
-      - name: Test jar
-        run: |
-          javac -cp target/regorus-java-0.2.2.jar Test.java
-          java -Djava.library.path=../../target/release -cp target/regorus-java-0.2.2.jar:. Test
-        working-directory: ./bindings/java
+      - name: Run Java smoke tests via xtask
+        run: cargo xtask test-java --release --frozen

.github/workflows/test-musl.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: musl
 
 on:
@@ -5,6 +7,9 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 env:
   CARGO_TERM_COLOR: always
@@ -15,19 +20,19 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v3
-    - name: Add musl target
-      run: rustup target add x86_64-unknown-linux-musl
+    - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+    - uses: ./.github/actions/toolchains/rust
+      with:
+        targets: x86_64-unknown-linux-musl
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus
+    - name: Fetch dependencies
+      run: cargo fetch --locked
+    - name: Fetch MUSL target dependencies
+      run: cargo fetch --locked --target x86_64-unknown-linux-musl
     - name: Install musl-gcc
       run: sudo apt update && sudo apt install -y musl-tools
-    - name: Build (MUSL)
-      run: cargo build --verbose --all-targets --target x86_64-unknown-linux-musl
-    - name: Run tests (MUSL)
-      run: cargo test -r --verbose --target x86_64-unknown-linux-musl
-    - name: Run tests (MUSL ACI)
-      run: cargo test -r --test aci --target x86_64-unknown-linux-musl
-    - name: Run tests (KATA ACI)
-      run: cargo test -r --test kata --target x86_64-unknown-linux-musl
-    - name: Run tests (MUSL OPA Conformance)
-      run: >-
-        cargo test -r --test opa --features opa-testutil,serde_json/arbitrary_precision --target x86_64-unknown-linux-musl  -- $(tr '\n' ' ' < tests/opa.passing)
+    - name: Run MUSL suite via xtask
+      run: cargo xtask test-musl --release --frozen --target x86_64-unknown-linux-musl

.github/workflows/test-no-std.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/no-std
 
 on:
@@ -5,6 +7,9 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 env:
   CARGO_TERM_COLOR: always
@@ -15,10 +20,18 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v3
-    - name: Add no_std target
-      run: rustup target add thumbv7m-none-eabi
-    - name: Build
-      run: cargo build -r --target thumbv7m-none-eabi
-      working-directory: ./tests/ensure_no_std
+    - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+    - uses: ./.github/actions/toolchains/rust
+      with:
+        targets: thumbv7m-none-eabi
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus
+    - name: Fetch dependencies
+      run: cargo fetch --locked
+    - name: Fetch ensure_no_std crate dependencies
+      run: cargo fetch --locked --manifest-path tests/ensure_no_std/Cargo.toml --target thumbv7m-none-eabi
+    - name: Test no-std
+      run: cargo xtask test-no-std --release --frozen
 

.github/workflows/test-python.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/python
 
 on:
@@ -5,34 +7,85 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
-env:
-  PYTHON_VERSION: "3.10"
-    
 jobs:
-  test:
-    runs-on: ubuntu-latest
+  build:
+    strategy:
+      matrix:
+        host:
+          - name: ubuntu-22.04
+            target: x86_64-unknown-linux-gnu
+          - name: windows-latest
+            target: x86_64-pc-windows-msvc
+    runs-on: ${{ matrix.host.name }}
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
-
-      - uses: actions/setup-python@v4
+      - uses: ./.github/actions/toolchains/rust
         with:
-          python-version: ${{ env.PYTHON_VERSION }}
+          targets: ${{ matrix.host.target }}
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch Python crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/python/Cargo.toml --target ${{ matrix.host.target }}
+
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.10"
           architecture: x64
 
-      - name: Build wheels
-        uses: PyO3/maturin-action@63b75c597b83e247fbf4fb7719801cc4220ae9f3 # v1.43.0
-        with:
-          target: x86_64
-          args: --release --out dist --manifest-path bindings/python/Cargo.toml
-          sccache: 'true'
+      - name: Install maturin
+        run: python -m pip install maturin==1.5.1
 
-      - name: Test wheel
-        run:  |
-          pip3 install dist/regorus-*.whl
-          cd bindings/python
-          cargo clippy --all-targets --no-deps -- -Dwarnings
-          python3 test.py          
+      - name: Build Python wheel via xtask
+        run: cargo xtask build-python --release --target ${{ matrix.host.target }} --target-dir bindings/python/dist --frozen
+
+      - name: Upload wheel artefacts
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
+        with:
+          name: regorus-wheel-${{ matrix.host.name }}
+          path: bindings/python/dist/regorus-*.whl
+
+  test:
+    needs: build
+    strategy:
+      matrix:
+        host: [ubuntu-24.04, ubuntu-22.04, windows-latest]
+        python-version: ["3.10", "3.11", "3.12", "3.13"]
+    runs-on: ${{ matrix.host }}
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          fetch-depth: 0
+      - uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch Python crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/python/Cargo.toml
+
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: ${{ matrix.python-version }}
+          architecture: x64
+
+      - name: Install maturin
+        run: python -m pip install maturin==1.5.1
+
+      - name: Run Python smoke tests via xtask
+        run: cargo xtask test-python --release --python python

.github/workflows/test-ruby.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/ruby
 
 on:
@@ -8,24 +10,35 @@ on:
 
 jobs:
   test:
+    if: false # temporarily disabled
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
 
       - name: Setup Ruby and Rust
-        uses: oxidize-rb/actions/setup-ruby-and-rust@7ca44a16e287e5ff7dd72ab53f4bd41cbf34a571 #v1.26
+        uses: oxidize-rb/actions/setup-ruby-and-rust@e5f9a49a7812a078584072f6e3f657ad247c8771 # v1.26
         with:
-          ruby-version: "3.3.1"
+          bundler: 2.6.5
+          rubygems: 3.6.5
+          ruby-version: "3.4.2"
           rustup-toolchain: "stable"
           bundler-cache: true
           cargo-cache: true
           working-directory: "bindings/ruby"
 
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+        with:
+          shared-key: ${{ runner.os }}-regorus
+
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch Ruby crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/ruby/Cargo.toml
+
       - name: Run ruby tests
-        run:  |
-          cd bindings/ruby
-          cargo clippy --all-targets --no-deps -- -Dwarnings
-          bundle exec rake
+        run: cargo xtask test-ruby --release --frozen

.github/workflows/test-wasm.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: bindings/wasm
 
 on:
@@ -5,28 +7,38 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 jobs:
   test:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0
 
-      - name: Setup Node
-        uses: actions/setup-node@v4
+      - name: Setup Rust toolchain
+        uses: ./.github/actions/toolchains/rust
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
         with:
-          node-version: 18
+          shared-key: ${{ runner.os }}-regorus
+      - name: Fetch dependencies
+        run: cargo fetch --locked
+
+      - name: Fetch WASM crate dependencies
+        run: cargo fetch --locked --manifest-path bindings/wasm/Cargo.toml
+
+      - name: Setup Node
+        uses: actions/setup-node@53b83947a5a98c8d113130e565377fae1a50d02f # v6.3.0
+        with:
+          node-version: 22
 
       - name: Install wasmlpack
         run: cargo install wasm-pack
 
-      - name: Test wasm binding
-        run: |
-          cd bindings/wasm
-          cargo clippy --all-targets --no-deps -- -Dwarnings
-          wasm-pack build --target nodejs --release
-          wasm-pack test --release --node
-          node test.js
+      - name: Test wasm binding via xtask
+        run: cargo xtask test-wasm --release --frozen --node node

.github/workflows/tests-debug.yml

@@ -1,3 +1,5 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+#
 name: tests/debug
 
 on:
@@ -5,6 +7,9 @@ on:
     branches: [ "main" ]
   pull_request:
     branches: [ "main" ]
+  schedule:
+    # Run at 8:00 AM every day
+    - cron: "0 8 * * *"
 
 env:
   CARGO_TERM_COLOR: always
@@ -15,23 +20,14 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v4
-    - name: Build (all features)
-      run: cargo build --all-features
-    - name: Build
-      run: cargo build
-    - name: Test no_std
-      run: cargo test --no-default-features
-    - name: Build only std
-      run: cargo build --example regorus --no-default-features --features "std"
-    - name: Doc Tests
-      run: cargo test --doc
-    - name: Run tests
-      run: cargo test
-    - name: Run tests (ACI)
-      run: cargo test --test aci
-    - name: Run tests (KATA)
-      run: cargo test --test kata
-    - name: Run tests (OPA Conformance)
-      run: >-
-        cargo test --test opa --features opa-testutil,serde_json/arbitrary_precision  -- $(tr '\n' ' ' < tests/opa.passing)
+    - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+    - name: Setup Rust toolchain
+      uses: ./.github/actions/toolchains/rust
+    - name: Cache cargo
+      uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
+      with:
+        shared-key: ${{ runner.os }}-regorus
+    - name: Fetch dependencies
+      run: cargo fetch --locked
+    - name: Run debug CI suite
+      run: cargo xtask ci-debug --frozen

.gitignore

@@ -4,10 +4,6 @@
 **/wheels/
 **/__pycache__/
 
-# Remove Cargo.lock from gitignore if creating an executable, leave it for libraries
-# More information here https://doc.rust-lang.org/cargo/guide/cargo-toml-vs-cargo-lock.html
-Cargo.lock
-
 # These are backup files generated by rustfmt
 **/*.rs.bk
 
@@ -25,4 +21,39 @@ worktrees/
 
 # Generated C, C++ headers
 bindings/ffi/regorus.h
-bindings/ffi/regorus.ffi.hpp
+bindings/ffi/regorus.ffi.hpp
+
+bindings/*/target
+
+# Temporary commit message files
+.commit-msg.txt
+
+# Local planning docs
+docs/plans/
+
+# C# build folders
+**bin
+**obj
+bindings/csharp/.nuget/
+
+# Bundler binstubs regenerated during ruby setup
+bindings/ruby/bin/
+
+# Visual Studio folders
+**/*.vs
+
+# Visual Studio solution files
+*.sln
+
+# JetBrains IDEs files
+.idea/
+
+# Java build artifacts
+**/*.class
+**/*.jar
+bindings/java/.classpath
+bindings/java/.project
+bindings/java/.settings/
+
+# Emacs temporary files
+*~

CHANGELOG.md

@@ -6,6 +6,114 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+- Azure RBAC condition interpreter with builtin evaluation coverage and YAML test suite, including quantifier (ForAnyOfAnyValues/ForAllOfAllValues), datetime (DateTimeEquals), IP (IpInRange), GUID (GuidEquals), list (ListContains), and string (StringEquals) semantics.
+- FFI surface for Azure RBAC condition evaluation (see bindings changelog for language-specific wrappers).
+
+### Changed
+- [**breaking**] Switch RVM binary serialization to postcard, bump the format to v4, and mark v1-3 loads as partial (recompile required).
+
+## [0.9.1](https://github.com/microsoft/regorus/compare/regorus-v0.9.0...regorus-v0.9.1) - 2026-02-06
+
+### Fixed
+- Release native C# handles reliably to avoid memory growth ([#571](https://github.com/microsoft/regorus/pull/571)).
+- Centralize C# handle gating with a short dispose wait and deferred release to avoid leaks while blocking new calls ([#571](https://github.com/microsoft/regorus/pull/571)).
+
+### Added
+- Manual C# memory growth tests for both `using` and finalizer paths ([#571](https://github.com/microsoft/regorus/pull/571)).
+- C# test runner options for filtered tests, console logging, and skipping sample apps ([#571](https://github.com/microsoft/regorus/pull/571)).
+
+## [0.5.0](https://github.com/microsoft/regorus/compare/regorus-v0.4.0...regorus-v0.5.0) - 2025-07-08
+
+### Added
+
+- [**breaking**] Indexes for nodes in the AST ([#414](https://github.com/anakrish/regorus/pull/414))
+- Updates for Policy Framework ([#405](https://github.com/anakrish/regorus/pull/405))
+- Regorus nuget package ([#383](https://github.com/anakrish/regorus/pull/383))
+
+### Fixed
+
+- emit import warning to stderr ([#430](https://github.com/anakrish/regorus/pull/430))
+- Clippy warnings ([#424](https://github.com/anakrish/regorus/pull/424))
+- Disallow else blocks for set rules ([#403](https://github.com/anakrish/regorus/pull/403))
+- [**breaking**] Remove cryptographic builtins ([#396](https://github.com/anakrish/regorus/pull/396))
+- [**breaking**] Fix glob.match behavior in presence of : ([#390](https://github.com/anakrish/regorus/pull/390))
+- C# EvalRule ([#387](https://github.com/anakrish/regorus/pull/387))
+
+### Other
+
+- Update release-plz action to v0.5.108 ([#431](https://github.com/anakrish/regorus/pull/431))
+- Early return for 'some in' statement ([#427](https://github.com/anakrish/regorus/pull/427))
+- Support manually generating C# bindings via Github action and add a README ([#423](https://github.com/anakrish/regorus/pull/423))
+- Make the bindings/cpp CMake project installable ([#416](https://github.com/anakrish/regorus/pull/416))
+- *(deps)* bump clap from 4.5.38 to 4.5.39 ([#415](https://github.com/anakrish/regorus/pull/415))
+- *(deps)* Update criterion and other deps ([#412](https://github.com/anakrish/regorus/pull/412))
+- Basic benchmarking setup with Criterion ([#408](https://github.com/anakrish/regorus/pull/408))
+- Default to Rego v1 in `regorus parse` ([#407](https://github.com/anakrish/regorus/pull/407))
+- *(deps)* bump clap from 4.5.37 to 4.5.38 ([#406](https://github.com/anakrish/regorus/pull/406))
+- Update dependencies ([#401](https://github.com/anakrish/regorus/pull/401))
+- Add C# test examples ([#397](https://github.com/anakrish/regorus/pull/397))
+- *(deps)* bump clap from 4.5.35 to 4.5.36 ([#395](https://github.com/anakrish/regorus/pull/395))
+- Python binding portability ([#388](https://github.com/anakrish/regorus/pull/388))
+- *(deps)* bump clap from 4.5.34 to 4.5.35 ([#389](https://github.com/anakrish/regorus/pull/389))
+- Use VersionPrefix and VersionSuffix ([#385](https://github.com/anakrish/regorus/pull/385))
+- Check-in Cargo.lock files and lockdown .net ([#384](https://github.com/anakrish/regorus/pull/384))
+
+## [0.4.0](https://github.com/microsoft/regorus/compare/regorus-v0.3.0...regorus-v0.4.0) - 2025-03-14
+
+### Fixed
+- [**breaking**] Update ruby json dependency ([#381](https://github.com/microsoft/regorus/pull/381))
+- [**breaking**] Remove ring dependency ([#380](https://github.com/microsoft/regorus/pull/380))
+- [**breaking**] Remove sha1 dependency ([#379](https://github.com/microsoft/regorus/pull/379))
+
+### Other
+- Specify optimization flags ([#378](https://github.com/microsoft/regorus/pull/378))
+
+## [0.3.0](https://github.com/microsoft/regorus/compare/regorus-v0.2.8...regorus-v0.3.0) - 2025-03-10
+
+### Added
+- [**breaking**] Update to OPA v1.2.0 ([#373](https://github.com/microsoft/regorus/pull/373))
+
+### Other
+- *(deps)* update pyo3 requirement from 0.23.5 to 0.24.0 ([#375](https://github.com/microsoft/regorus/pull/375))
+- Update ruby binding deps, ruby gem version 0.2.3 ([#374](https://github.com/microsoft/regorus/pull/374))
+- *(deps)* update pyo3 requirement from 0.22.0 to 0.23.5 ([#372](https://github.com/microsoft/regorus/pull/372))
+- *(deps)* update rand requirement from 0.8.5 to 0.9.0 ([#370](https://github.com/microsoft/regorus/pull/370))
+- *(deps)* update cbindgen requirement from 0.27.0 to 0.28.0 ([#361](https://github.com/microsoft/regorus/pull/361))
+- Fix typo in README.md ([#366](https://github.com/microsoft/regorus/pull/366))
+- Update dependencies ([#369](https://github.com/microsoft/regorus/pull/369))
+- Fix clippy warning for result? ([#362](https://github.com/microsoft/regorus/pull/362))
+- *(deps)* update itertools requirement from 0.13.0 to 0.14.0 ([#357](https://github.com/microsoft/regorus/pull/357))
+- *(deps)* update jsonschema requirement from 0.26.1 to 0.28.1 ([#356](https://github.com/microsoft/regorus/pull/356))
+- resolve anyhow compile errors ([#355](https://github.com/microsoft/regorus/pull/355))
+- *(deps)* update prettydiff requirement from 0.7.0 to 0.8.0 ([#348](https://github.com/microsoft/regorus/pull/348))
+
+## [0.2.8](https://github.com/microsoft/regorus/compare/regorus-v0.2.7...regorus-v0.2.8) - 2024-11-06
+
+### Other
+- *(deps)* update jsonschema requirement from 0.24.0 to 0.26.1 ([#343](https://github.com/microsoft/regorus/pull/343))
+- Update to OPA v0.70.0 ([#341](https://github.com/microsoft/regorus/pull/341))
+
+## [0.2.7](https://github.com/microsoft/regorus/compare/regorus-v0.2.6...regorus-v0.2.7) - 2024-10-22
+
+### Fixed
+- docs failing to build ([#334](https://github.com/microsoft/regorus/pull/334))
+
+### Other
+- *(deps)* update jsonschema requirement from 0.23.0 to 0.24.0 ([#332](https://github.com/microsoft/regorus/pull/332))
+- *(deps)* update jsonschema requirement from 0.22.3 to 0.23.0 ([#331](https://github.com/microsoft/regorus/pull/331))
+
+## [0.2.6](https://github.com/microsoft/regorus/compare/regorus-v0.2.5...regorus-v0.2.6) - 2024-10-09
+
+### Added
+- integer conversion functions for Value ([#328](https://github.com/microsoft/regorus/pull/328))
+
+### Other
+- update to OPA v0.69.0 ([#327](https://github.com/microsoft/regorus/pull/327))
+- *(deps)* update jsonschema requirement from 0.21.0 to 0.22.3 ([#326](https://github.com/microsoft/regorus/pull/326))
+- *(deps)* update jsonschema requirement from 0.20.0 to 0.21.0 ([#325](https://github.com/microsoft/regorus/pull/325))
+- update to jsonschema 0.20.0 ([#323](https://github.com/microsoft/regorus/pull/323))
+
 ## [0.2.5](https://github.com/microsoft/regorus/compare/regorus-v0.2.4...regorus-v0.2.5) - 2024-09-18
 
 ### Added

Cargo.toml

@@ -1,49 +1,54 @@
 [workspace]
 
 members = [
-    "bindings/ffi",
-    "bindings/python",
-    "bindings/wasm",
-    "bindings/java",
-    "bindings/ruby/ext/regorusrb",
     "tests/ensure_no_std",
+  "xtask",
 ]
 
 [package]
 name = "regorus"
 description = "A fast, lightweight Rego (OPA policy language) interpreter"
-version = "0.2.5"
+version = "0.9.1"
 edition = "2021"
-license-file = "LICENSE"
+license = "MIT AND Apache-2.0 AND BSD-3-Clause"
 repository = "https://github.com/microsoft/regorus"
 keywords = ["interpreter", "no_std", "opa", "policy-as-code", "rego"]
 
+# Enable verification with Verus
+
+[package.metadata.verus]
+verify = true
+
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [lib]
 doctest = false
 
 [features]
-default = ["full-opa", "arc"]
+default = ["full-opa", "arc", "rvm"]
 
-arc = ["scientific/arc"]
+arc = []
 ast = []
+azure_policy = ["dep:jsonschema", "dep:chrono", "dep:ipnet", "dep:icu_casemap", "dep:hashbrown", "arc", "dashmap"]
+azure-rbac = ["regex", "time", "net"]
 base64 = ["dep:data-encoding"]
 base64url = ["dep:data-encoding"]
 coverage = []
-crypto = ["dep:constant_time_eq", "dep:hmac", "dep:hex", "dep:md-5", "dep:sha1", "dep:sha2"]
-deprecated = []
 hex = ["dep:data-encoding"]
 http = []
-glob = ["dep:wax"]
+glob = ["dep:globset"]
 graph = []
 jsonschema = ["dep:jsonschema"]
-jwt = ["dep:jsonwebtoken", "dep:data-encoding", "dep:itertools"]
+mimalloc = ["dep:mimalloc"]
+net = ["dep:ipnet"]
 no_std = ["lazy_static/spin_no_std"]
 opa-runtime = []
 regex = ["dep:regex"]
+cache = ["dep:lru"]
+rvm = ["dep:postcard", "dep:indexmap"]
 semver = ["dep:semver"]
-std = ["rand/std", "rand/std_rng", "serde_json/std"]
+allocator-memory-limits = ["std", "mimalloc", "mimalloc/allocator-memory-limits"]
+std = ["rand/std", "rand/std_rng", "serde_json/std", "indexmap?/std", "msvc_spectre_libs", "dep:parking_lot" ]
 time = ["dep:chrono", "dep:chrono-tz"]
 uuid = ["dep:uuid"]
 urlquery = ["dep:url"]
@@ -52,16 +57,15 @@ full-opa = [
     "base64",
     "base64url",
     "coverage",
-    "crypto",
-    "deprecated",
     "glob",
     "graph",
     "hex",
     "http",
-    "jwt",
     "jsonschema",
+    "net",
     "opa-runtime",
     "regex",
+    "cache",
     "semver",
     "std",
     "time",
@@ -79,8 +83,6 @@ opa-no-std = [
   "base64",
   "base64url",
   "coverage",
-  "crypto",
-  "deprecated",
   "graph",
   "hex",
   "no_std",
@@ -99,50 +101,66 @@ opa-testutil = []
 rand = ["dep:rand"]
 
 [dependencies]
-anyhow = { version = "1.0.45", default-features = false }
-serde = {version = "1.0.150", default-features = false, features = ["derive", "rc"] }
+anyhow = { version = "1.0.102", default-features = false }
+serde = {version = "1.0.150", default-features = false, features = ["derive", "rc", "alloc"] }
 serde_json = { version = "1.0.89", default-features = false, features = ["alloc"] }
+hashbrown = { version = "0.16", default-features = false, features = ["default-hasher"], optional = true }
 lazy_static =  { version = "1.4.0", default-features = false }
+thiserror = { version = "2.0", default-features = false }
 
-# Crypto
-constant_time_eq = {version = "0.3.0", optional = true, default-features = false }
-hmac = {version = "0.12.1", optional = true, default-features = false}
-sha2 = {version= "0.10.8", optional = true, default-features = false }
-hex = {version = "0.4.3", optional = true, default-features = false, features = ["alloc"] }
-sha1 = {version = "0.10.6", optional = true, default-features = false }
-md-5 = {version = "0.10.6", optional = true, default-features = false }
+data-encoding = { version = "2.8.0", optional = true, default-features=false, features = ["alloc"] }
+num-bigint = { version = "0.4", default-features = false }
+num-traits = { version = "0.2", default-features = false }
+parking_lot = { version = "0.12", optional = true }
+spin = { version = "0.10.0", default-features = false, features = ["mutex", "spin_mutex"] }
 
-data-encoding = { version = "2.4.0", optional = true, default-features=false, features = ["alloc"] }
-scientific = { version = "0.5.2" }
-
-regex = {version = "1.10.2", optional = true, default-features = false }
-semver = {version = "1.0.20", optional = true, default-features = false }
-wax = { version = "0.6.0", features = [], default-features = false, optional = true }
-url = { version = "2.5.0", optional = true }
-uuid = { version = "1.6.1", default-features = false, features = ["v4", "fast-rng"], optional = true }
-jsonschema = { version = "0.19.1", default-features = false, optional = true }
-chrono = { version = "0.4.31", optional = true }
-chrono-tz = { version = "0.10.0", optional = true }
-jsonwebtoken = { version = "9.2.0", optional = true }
-itertools = { version = "0.13.0", default-features = false, optional = true }
+globset = { version = "0.4.16", features = ["simd-accel"], default-features = false, optional = true }
+regex = {version = "1.12.3", optional = true, default-features = false }
+semver = {version = "1.0.25", optional = true, default-features = false }
+url = { version = "2.5.4", optional = true }
+uuid = { version = "1.22.0", default-features = false, features = ["v4", "fast-rng"], optional = true }
+jsonschema = { version = "0.45.0", default-features = false, optional = true }
+chrono = { version = "0.4.44", optional = true }
+chrono-tz = { version = "0.10.1", optional = true }
+ipnet = { version = "2.12.0", optional = true, default-features = false }
+icu_casemap = { version = "2.1", optional = true, default-features = false, features = ["compiled_data"] }
 
 serde_yaml = {version = "0.9.16", default-features = false, optional = true }
-rand = { version = "0.8.5", default-features = false, optional = true }
+# Specify thread_rng for in order to use random_range
+rand = { version = "0.10.0", default-features = false, features = ["thread_rng"], optional = true }
+
+# Causes the project to link with the Spectre-mitigated CRT and libs.
+msvc_spectre_libs = { version = "0.1", features = ["error"], optional = true }
+dashmap = { version = "6.1", default-features = false, optional = true }
+lru = { version = "0.16", default-features = false, optional = true }
+mimalloc = { package = "regorus-mimalloc", path = "mimalloc", version = "2.2.6", optional = true }
+
+# rvm related deps
+indexmap = { version = "2.12.1", default-features = false, features = ["serde"], optional = true }
+postcard = { version = "1.1.3", default-features = false, features = ["alloc"], optional = true }
+
+# Use Verus for verification
+vstd = { version = "0.0.0-2026-03-17-2326" }
 
 [dev-dependencies]
-anyhow = "1.0.45"
+anyhow = "1.0.102"
 cfg-if = "1.0.0"
-clap = { version = "4.4.7", features = ["derive"] }
-prettydiff = { version = "0.7.0", default-features = false }
+clap = { version = "4.5.53", features = ["derive"] }
+prettydiff = { version = "0.9.0", default-features = false }
 serde_yaml = "0.9.16"
 test-generator = "0.3.1"
 walkdir = "2.3.2"
+criterion = { version = "0.8" }
+
+num_cpus = "1.16"
 
 [build-dependencies]
 anyhow = "1.0"
 
 [profile.release]
 debug = true
+lto = true
+codegen-units = 1
 
 [[test]]
 name="opa"
@@ -160,6 +178,39 @@ name="kata"
 harness=false
 test=false
 
+[[bench]]
+name = "regorus_benchmark"
+harness = false
+
+[[bench]]
+name = "schema_validation_benchmark"
+harness = false
+required-features = ["azure_policy"]
+
+[[bench]]
+name = "engine_evaluation_benchmark"
+path = "benches/evaluation/engine_evaluation_benchmark.rs"
+harness = false
+
+[[bench]]
+name = "compiled_policy_evaluation_benchmark"
+path = "benches/evaluation/compiled_policy_evaluation_benchmark.rs"
+harness = false
+
+[[bench]]
+name = "aci_benchmark"
+harness = false
+
+[[bench]]
+name = "rvm_benchmark"
+harness = false
+required-features = ["rvm"]
+
+[[bench]]
+name = "normalization_benchmark"
+harness = false
+required-features = ["azure_policy"]
+
 [[example]]
 name="regorus"
 harness=false
@@ -171,3 +222,6 @@ doctest=false
 # RUSTDOCFLAGS="--cfg docsrs" cargo +nightly doc --all-features --no-deps
 all-features = true
 rustdoc-args = ["--cfg", "docsrs"]
+
+[lints.rust]
+unexpected_cfgs = { level = "warn", check-cfg = ['cfg(verus_keep_ghost)'] }

LICENSE

@@ -19,3 +19,238 @@
     LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
     OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
     SOFTWARE
+
+The file src/builtins/time/diff.rs contains code derived from Go's `time`
+package, which carries the following license:
+
+    Copyright (c) 2009 The Go Authors. All rights reserved.
+    
+    Redistribution and use in source and binary forms, with or without
+    modification, are permitted provided that the following conditions are
+    met:
+    
+       * Redistributions of source code must retain the above copyright
+    notice, this list of conditions and the following disclaimer.
+       * Redistributions in binary form must reproduce the above
+    copyright notice, this list of conditions and the following disclaimer
+    in the documentation and/or other materials provided with the
+    distribution.
+       * Neither the name of Google Inc. nor the names of its
+    contributors may be used to endorse or promote products derived from
+    this software without specific prior written permission.
+    
+    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+    "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+    A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+    OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+    SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+    LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+    DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+    THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+    (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+Some files are licensed Apache-2.0 (LICENSE-2.0.txt).
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -3,14 +3,14 @@
 **Regorus** is
 
   - *Rego*-*Rus(t)*  - A fast, light-weight [Rego](https://www.openpolicyagent.org/docs/latest/policy-language/)
-   interpreter written in Rust.
+    interpreter written in Rust.
   - *Rigorous* - A rigorous enforcer of well-defined Rego semantics.
 
 Regorus is also
   - *cross-platform* - Written in platform-agnostic Rust.
   - *no_std compatible* - Regorus can be used in `no_std` environments too. Most of the builtins are supported.
-  - *current* - We strive to keep Regorus up to date with latest OPA release. Regorus supports `import rego.v1`.
-  - *compliant* - Regorus is mostly compliant with the latest [OPA release v0.68.0](https://github.com/open-policy-agent/opa/releases/tag/v0.68.0). See [OPA Conformance](#opa-conformance) for details. Note that while we behaviorally produce the same results, we don't yet support all the builtins.
+  - *current* - We strive to keep Regorus up to date with latest OPA release. Regorus defaults to `v1` of the Rego language.
+  - *compliant* - Regorus is mostly compliant with the latest [OPA release v1.2.0](https://github.com/open-policy-agent/opa/releases/tag/v1.2.0). See [OPA Conformance](#opa-conformance) for details. Note that while we behaviorally produce the same results, we don't yet support all the builtins.
   - *extensible* - Extend the Rego language by implementing custom stateful builtins in Rust.
     See [add_extension](https://github.com/microsoft/regorus/blob/fc68bf9c8bea36427dae9401a7d1f6ada771f7ab/src/engine.rs#L352).
     Support for extensibility using other languages coming soon.
@@ -32,7 +32,6 @@ fn main() -> anyhow::Result<()> {
     let policy = String::from(
         r#"
        package example
-       import rego.v1
 
        allow if {
           ## All actions are allowed for admins.
@@ -99,7 +98,7 @@ $ cargo build -r --example regorus --no-default-features; strip target/release/e
 -rwxr-xr-x  1 anand  staff   1.9M May 11 22:04 target/release/examples/regorus*
 ```
 
-Regorus passes the [OPA v0.68.0 test-suite](https://www.openpolicyagent.org/docs/latest/ir/#test-suite) barring a few
+Regorus passes the [OPA v1.2.0 test-suite](https://www.openpolicyagent.org/docs/latest/ir/#test-suite) barring a few
 builtins. See [OPA Conformance](#opa-conformance) below.
 
 ## Bindings
@@ -108,11 +107,11 @@ Regorus can be used from a variety of languages:
 
 - *C*: C binding is generated using [cbindgen](https://github.com/mozilla/cbindgen).
   [corrosion-rs](https://github.com/corrosion-rs/corrosion) can be used to seamlessly use Regorous
-   in your CMake based projects. See [bindings/c](https://github.com/microsoft/regorus/tree/main/bindings/c).
+  in your CMake based projects. See [bindings/c](https://github.com/microsoft/regorus/tree/main/bindings/c).
 - *C freestanding*: [bindings/c_no_std](https://github.com/microsoft/regorus/tree/main/bindings/c_no_std) shows how to use Regorus from C environments without a libc.
 - *C++*: C++ binding is generated using [cbindgen](https://github.com/mozilla/cbindgen).
   [corrosion-rs](https://github.com/corrosion-rs/corrosion) can be used to seamlessly use Regorous
-   in your CMake based projects. See [bindings/cpp](https://github.com/microsoft/regorus/tree/main/bindings/cpp).
+  in your CMake based projects. See [bindings/cpp](https://github.com/microsoft/regorus/tree/main/bindings/cpp).
 - *C#*: C# binding is generated using [csbindgen](https://github.com/Cysharp/csbindgen). See [bindings/csharp](https://github.com/microsoft/regorus/tree/main/bindings/csharp) for an example of how to build and use Regorus in your C# projects.
 - *Golang*: The C bindings are exposed to Golang via [CGo](https://pkg.go.dev/cmd/cgo). See [bindings/go](https://github.com/microsoft/regorus/tree/main/bindings/go) for an example of how to build and use Regorus in your Go projects.
 - *Python*: Python bindings are generated using [pyo3](https://github.com/PyO3/pyo3). Wheels are created using [maturin](https://github.com/PyO3/maturin). See [bindings/python](https://github.com/microsoft/regorus/tree/main/bindings/python).
@@ -146,6 +145,7 @@ $ regorus
 Usage: regorus <COMMAND>
 
 Commands:
+  ast    Parse a Rego policy and dump AST
   eval   Evaluate a Rego Query
   lex    Tokenize a Rego policy
   parse  Parse a Rego policy
@@ -274,9 +274,22 @@ Benchmark 1: opa eval -b tests/aci -d tests/aci/data.json -i tests/aci/input.jso
   Range (min … max):    43.8 ms …  46.7 ms    62 runs
 
 ```
+
+## Contributor Workflow
+
+Regorus uses a small companion CLI under the `xtask` package to keep CI and local development in sync.
+The commands mirror our GitHub Actions jobs, making it easy to dry-run CI steps before sending a pull request.
+
+- Run the full release pipeline with `cargo xtask ci-release` and the debug checks with `cargo xtask ci-debug`.
+- Exercise language bindings through focused helpers such as `cargo xtask test-java --release --frozen` or `cargo xtask test-go`.
+- Use `cargo xtask test-musl --release --frozen` for the cross-compilation matrix and `cargo xtask test-no-std` for embedded targets.
+- Formatting (`cargo xtask fmt`) and linting (`cargo xtask clippy --sarif`) wrap the usual Cargo tooling while matching CI defaults.
+
+The workflows in `.github/workflows` invoke the same commands, so keeping local runs green is usually enough to satisfy the checks enforced on `main`.
+
 ## OPA Conformance
 
-Regorus has been verified to be compliant with [OPA v0.68.0](https://github.com/open-policy-agent/opa/releases/tag/v0.68.0)
+Regorus has been verified to be compliant with [OPA v1.2.0](https://github.com/open-policy-agent/opa/releases/tag/v1.2.0)
 using a [test driver](https://github.com/microsoft/regorus/blob/main/tests/opa.rs) that loads and runs the OPA testsuite using Regorus, and verifies that expected outputs are produced.
 
 The test driver can be invoked by running:
@@ -288,29 +301,23 @@ $ cargo test -r --test opa --features opa-testutil,serde_json/arbitrary_precisio
 Currently, Regorus passes all the non-builtin specific tests.
 See [passing tests suites](https://github.com/microsoft/regorus/blob/main/tests/opa.passing).
 
-The following test suites don't pass fully due to mising builtins:
-- `cryptoparsersaprivatekeys`
-- `cryptox509parseandverifycertificates`
-- `cryptox509parsecertificaterequest`
-- `cryptox509parsecertificates`
-- `cryptox509parsekeypair`
-- `cryptox509parsersaprivatekey`
+The following test suites don't pass fully due to missing builtins:
 - `globsmatch`
 - `graphql`
 - `invalidkeyerror`
 - `jsonpatch`
+- `jwtbuiltins`
 - `jwtdecodeverify`
 - `jwtencodesign`
+- `jwtencodesignheadererrors`
+- `jwtencodesignpayloaderrors`
 - `jwtencodesignraw`
 - `jwtverifyhs256`
 - `jwtverifyhs384`
 - `jwtverifyhs512`
 - `jwtverifyrsa`
-- `netcidrcontains`
 - `netcidrcontainsmatches`
-- `netcidrexpand`
 - `netcidrintersects`
-- `netcidrisvalid`
 - `netcidrmerge`
 - `netcidroverlap`
 - `netlookupipaddr`
@@ -322,6 +329,7 @@ The following test suites don't pass fully due to mising builtins:
 
 They are captured in the following [github issues](https://github.com/microsoft/regorus/issues?q=is%3Aopen+is%3Aissue+label%3Alib).
 
+Cryptographic builtins are not supported by design. Users that need cryptographic builtins are encouraged to use [extensions](https://docs.rs/regorus/latest/regorus/struct.Engine.html#method.add_extension).
 
 ### Grammar
 

benches/aci_benchmark.rs

@@ -0,0 +1,80 @@
+// Copyright (c) Microsoft Corporation.
+// Licensed under the MIT License.
+use regorus::{Engine, Value};
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion};
+use serde::{Deserialize, Serialize};
+use walkdir::WalkDir;
+
+use std::path::Path;
+
+#[derive(Serialize, Deserialize, PartialEq, Debug)]
+struct TestCase {
+    note: String,
+    data: Value,
+    input: Value,
+    modules: Vec<String>,
+    query: String,
+    want_result: Value,
+}
+
+#[derive(Serialize, Deserialize, PartialEq, Debug)]
+struct YamlTest {
+    cases: Vec<TestCase>,
+}
+
+fn aci_policy_eval(c: &mut Criterion) {
+    let dir = Path::new("tests/aci");
+    for entry in WalkDir::new(dir)
+        .sort_by_file_name()
+        .into_iter()
+        .filter_map(|e| e.ok())
+    {
+        let path = entry.path();
+        if !path.to_string_lossy().ends_with(".yaml") {
+            continue;
+        }
+
+        let yaml = std::fs::read(path).expect("failed to read yaml test");
+        let yaml = String::from_utf8_lossy(&yaml);
+        let test: YamlTest = serde_yaml::from_str(&yaml).expect("failed to deserialize yaml test");
+
+        for case in &test.cases {
+            let rule = case.query.replace("=x", "");
+            c.bench_with_input(
+                BenchmarkId::new("case ", format!("{} {}", &case.note, &rule)),
+                &case,
+                |b, case| {
+                    let mut engine = Engine::new();
+                    engine.set_rego_v0(true);
+
+                    engine
+                        .add_data(case.data.clone())
+                        .expect("failed to add data");
+                    engine.set_input(case.input.clone());
+
+                    for (idx, rego) in case.modules.iter().enumerate() {
+                        if rego.ends_with(".rego") {
+                            let path = dir.join(rego);
+                            let path = path.to_str().expect("not a valid path");
+                            engine
+                                .add_policy_from_file(path)
+                                .expect("failed to add policy");
+                        } else {
+                            engine
+                                .add_policy(format!("rego{idx}.rego"), rego.clone())
+                                .expect("failed to add policy");
+                        }
+                    }
+
+                    b.iter(|| {
+                        engine.eval_rule(rule.clone()).unwrap();
+                    })
+                },
+            );
+        }
+    }
+}
+
+criterion_group!(aci_benches, aci_policy_eval);
+criterion_main!(aci_benches);

benches/evaluation/README.md

@@ -0,0 +1,157 @@
+# Regorus Multi-Threaded Evaluation Benchmark
+
+A benchmark suite for measuring the multi-threaded performance of the Regorus policy evaluation engine.
+
+## Overview
+
+This benchmark evaluates the throughput and scalability of Regorus policy evaluation across different thread counts and configuration strategies. It measures performance variations between fresh and cloned engine instances, as well as fresh and cloned input data.
+
+## Features
+
+- **Multi-threaded evaluation** testing from 1 to `num_cpus * 2` threads
+- **Configurable engine strategies**: Fresh vs. cloned engine instances
+- **Configurable input strategies**: Fresh parsing vs. cloned input data
+- **Complex policy evaluation** using realistic RBAC and data sensitivity policies
+- **Criterion-based benchmarking** with statistical analysis
+- **Performance metrics** including throughput and timing
+
+## Benchmark Structure
+
+### Test Configurations
+
+The benchmark tests four different configuration combinations:
+
+1. **Cloned Engines + Cloned Inputs**: Pre-instantiated engines with pre-parsed input data
+2. **Cloned Engines + Fresh Inputs**: Pre-instantiated engines with fresh JSON parsing
+3. **Fresh Engines + Cloned Inputs**: New engine instances with pre-parsed input data
+4. **Fresh Engines + Fresh Inputs**: New engine instances with fresh JSON parsing
+
+### Thread Scaling
+
+Tests are performed with thread counts: 1, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32 (up to `num_cpus * 2`)
+
+Each thread performs 1000 policy evaluations to ensure statistically significant measurements.
+
+## Running the Benchmark
+
+### Prerequisites
+
+- Rust 1.70+
+- Cargo
+
+### Execution
+
+Run the complete benchmark suite:
+
+```bash
+cargo bench evaluation_benchmark
+```
+
+Run specific benchmarks:
+
+```bash
+# Run only cloned engines with cloned inputs
+cargo bench "cloned_engines , cloned_inputs"
+
+# Run only single-threaded tests
+cargo bench "1 threads"
+```
+
+### Output
+
+Results are generated in the `target/criterion/` directory and include:
+
+- Detailed timing statistics
+- Throughput measurements (Kelem/s)
+- Performance comparison with previous runs
+- HTML reports with graphs and analysis
+
+## Test Policies
+
+The benchmark uses complex Rego policies that simulate real-world scenarios:
+
+### RBAC Policy
+- Role-based access control with hierarchical permissions
+- User-role-resource mapping
+- Action-based authorization
+
+### Data Sensitivity Policy
+- Multi-level data classification (public, internal, confidential, secret)
+- Access level validation
+- Clearance-based filtering
+
+### Time-based Access Policy
+- Business hours validation
+- Temporal access control
+- Schedule-based permissions
+
+### Azure Resource Policies
+- **VM Deployment**: VM size restrictions, regional compliance, security configurations
+- **Storage Account Security**: Encryption requirements, network ACLs, HTTPS enforcement
+- **Key Vault Access**: Service principal validation, soft delete requirements, conditional access
+- **Network Security Groups**: Port restrictions, CIDR validation, priority-based rules
+
+### Policy Complexity Features
+- **Multi-condition validation**: Complex nested object property checks
+- **Network operations**: CIDR matching and IP range validation
+- **Time-based constraints**: Timestamp comparisons and business hour logic
+- **Security compliance**: Encryption, authentication, and access control patterns
+- **Azure Resource Manager**: Real-world cloud governance scenarios
+
+## Configuration
+
+### Benchmark Parameters
+
+- **Evaluations per thread**: 1000
+- **Measurement iterations**: 100 samples per configuration
+- **Warm-up time**: 3 seconds
+- **Measurement time**: 10 seconds (extended for high thread counts)
+
+### Customization
+
+The benchmark can be customized by modifying `evaluation_benchmark.rs`:
+
+```rust
+// Adjust evaluations per thread
+let evals_per_thread = 1000;
+
+// Modify thread count calculation
+let max_threads = num_cpus::get() * 2;
+
+// Configure test scenarios
+let scenarios = [
+    (true, true),   // cloned_engines, cloned_inputs
+    (true, false),  // cloned_engines, fresh_inputs
+    (false, true),  // fresh_engines, cloned_inputs
+    (false, false), // fresh_engines, fresh_inputs
+];
+```
+
+## Understanding Results
+
+### Metrics
+
+- **Total Evaluation Time**: Total execution time for all evaluations across all threads (ms)
+- **Throughput**: Evaluations per second measured in Kelem/s
+  - **Kelem/s**: Thousands of elements (policy evaluations) per second
+  - Example: 98.71 Kelem/s = 98,710 policy evaluations per second
+
+
+### Interpretation
+
+- **Lower time** = better performance
+- **Higher throughput** = better performance
+- **Consistent results** across runs indicate stable performance
+- **Outliers** may indicate system interference or measurement variance
+
+### Tips
+
+- Run on dedicated hardware for consistent results
+- Disable other applications during benchmarking
+- Use release builds for accurate performance measurements
+- Consider CPU affinity for highly controlled testing
+
+## Files
+
+- `evaluation_benchmark.rs`: Main benchmark implementation
+- Results are saved to `../../target/criterion/` directory

benches/evaluation/compiled_policy_evaluation_benchmark.md

@@ -0,0 +1,160 @@
+# Compiled Policy Evaluation Benchmark Results
+
+## Test Environment
+- **Platform**: Apple Silicon (M-Series)
+- **CPU**: 16 cores
+- **Architecture**: ARM64 (aarch64-apple-darwin)
+- **Rust Version**: 1.82.0
+- **Allocator**: mimalloc (default allocator)
+- **Benchmark Framework**: Criterion.rs
+- **Test Data**: 20,000 inputs per evaluation (1000 per thread)
+- **Policy**: Complex authorization policy with nested rules
+
+## Benchmark Overview
+
+The compiled policy evaluation benchmark tests Regorus compiled policy performance across multiple thread configurations (1-32 threads). It measures throughput (thousands of evaluations per second) for different combinations of compiled policy and input data reuse strategies.
+
+## Configuration Combinations
+
+1. **Compiled Shared Policies, Cloned Inputs**: Each thread uses shared compiled policies and clones of parsed input data - optimal for performance
+2. **Compiled Shared Policies, Fresh Inputs**: Each thread uses shared compiled policies but parses new inputs each time
+3. **Compiled Per Iteration, Cloned Inputs**: Each thread compiles the policy each iteration but reuses input data
+4. **Compiled Per Iteration, Fresh Inputs**: Each thread compiles new policies and parses new inputs for each iteration
+
+## Performance Results
+
+### Compiled Shared Policies, Cloned Inputs (Best Performance)
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                       2.35 |                  426 |
+|       2 |                       5.36 |                  373 |
+|       4 |                      11.70 |                  342 |
+|       6 |                      20.33 |                  295 |
+|       8 |                      43.26 |                  185 |
+|      10 |                      61.93 |                  162 |
+|      12 |                      79.30 |                  151 |
+|      14 |                      94.45 |                  148 |
+|      16 |                     113.39 |                  141 |
+|      18 |                     154.41 |                  117 |
+|      20 |                     184.37 |                  108 |
+|      22 |                     204.00 |                  108 |
+|      24 |                     220.45 |                  109 |
+|      26 |                     237.07 |                  110 |
+|      28 |                     252.58 |                  111 |
+|      30 |                     273.57 |                  110 |
+|      32 |                     292.69 |                  109 |
+
+### Compiled Shared Policies, Fresh Inputs
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                       3.34 |                  299 |
+|       2 |                       7.29 |                  274 |
+|       4 |                      15.19 |                  263 |
+|       6 |                      24.90 |                  241 |
+|       8 |                      49.22 |                  163 |
+|      10 |                      68.45 |                  146 |
+|      12 |                      86.55 |                  139 |
+|      14 |                     104.77 |                  134 |
+|      16 |                     136.07 |                  118 |
+|      18 |                     169.05 |                  106 |
+|      20 |                     198.25 |                  101 |
+|      22 |                     217.05 |                  101 |
+|      24 |                     234.75 |                  102 |
+|      26 |                     254.53 |                  102 |
+|      28 |                     276.06 |                  101 |
+|      30 |                     296.12 |                  101 |
+|      32 |                     318.81 |                  100 |
+
+### Compiled Per Iteration, Cloned Inputs
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                      18.11 |                   55 |
+|       2 |                      36.89 |                   54 |
+|       4 |                      75.46 |                   53 |
+|       6 |                     114.66 |                   52 |
+|       8 |                     152.80 |                   52 |
+|      10 |                     192.17 |                   52 |
+|      12 |                     232.32 |                   52 |
+|      14 |                     301.47 |                   46 |
+|      16 |                     380.36 |                   42 |
+|      18 |                     424.64 |                   42 |
+|      20 |                     484.76 |                   41 |
+|      22 |                     531.62 |                   41 |
+|      24 |                     582.88 |                   41 |
+|      26 |                     631.39 |                   41 |
+|      28 |                     671.99 |                   42 |
+|      30 |                     717.65 |                   42 |
+|      32 |                     766.05 |                   42 |
+
+### Compiled Per Iteration, Fresh Inputs
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                      19.07 |                   52 |
+|       2 |                      38.89 |                   51 |
+|       4 |                      79.52 |                   50 |
+|       6 |                     120.89 |                   50 |
+|       8 |                     161.08 |                   50 |
+|      10 |                     202.37 |                   49 |
+|      12 |                     244.04 |                   49 |
+|      14 |                     316.66 |                   44 |
+|      16 |                     398.02 |                   40 |
+|      18 |                     449.54 |                   40 |
+|      20 |                     500.57 |                   40 |
+|      22 |                     557.97 |                   39 |
+|      24 |                     605.71 |                   40 |
+|      26 |                     656.88 |                   40 |
+|      28 |                     710.03 |                   39 |
+|      30 |                     741.09 |                   40 |
+|      32 |                     801.26 |                   40 |
+
+## Analysis
+
+The compiled policy benchmark demonstrates the following performance characteristics with mimalloc as the default allocator:
+
+1. **Best Performance**: Compiled shared policies with cloned inputs provide the highest throughput
+2. **Compilation Impact**: 
+   - Pre-compiled policies: Significantly faster than per-iteration compilation
+   - Per-iteration compilation: Major overhead (~7-8x slower than pre-compiled)
+3. **Scaling Patterns with mimalloc**:
+   - Best throughput achieved at 1 thread for shared policy configurations
+   - mimalloc provides better thread scaling characteristics compared to the default allocator
+   - Higher thread counts show performance degradation due to contention, but less severe with mimalloc
+   - Per-iteration compilation shows poor scaling across all thread counts
+4. **Input Processing**: Fresh inputs add ~30% overhead across all configurations
+5. **Thread Performance with mimalloc**: 
+   - Peak performance at 1 thread for most configurations
+   - Reasonable performance maintained up to 12-16 threads for shared policies
+   - Compiled policies show better thread scaling than per-iteration compilation
+   - mimalloc helps reduce allocation-related contention in multi-threaded scenarios
+
+## Comparison with Engine Evaluation
+
+### Multi-Thread Performance Comparison
+
+| Configuration        | 1 Thread (Kelem/s) | 4 Threads (Kelem/s) | 8 Threads (Kelem/s) |
+|:---------------------|:-------------------|:--------------------|:--------------------|
+|                      | CP / EE            | CP / EE             | CP / EE             |
+| Shared/Cloned        | 426 / 423          | 342 / 406           | 185 / 341           |
+| Shared/Fresh         | 299 / 309          | 263 / 297           | 163 / 266           |
+| Per-iteration/Cloned | 55 / 56            | 53 / 54             | 52 / 53             |
+| Per-iteration/Fresh  | 52 / 53            | 50 / 51             | 50 / 51             |
+
+### Threading Efficiency Analysis
+
+| Configuration        | Low Contention (1-4t) | Medium Contention (6-12t) | High Contention (16+t) |
+|:---------------------|:----------------------|:--------------------------|:-----------------------|
+|                      | Avg CP / EE           | Avg CP / EE               | Avg CP / EE            |
+| Shared/Cloned        | 384 / 414             | 203 / 329                 | 123 / 250              |
+| Shared/Fresh         | 284 / 302             | 176 / 235                 | 108 / 201              |
+| Per-iteration/Cloned | 54 / 55               | 50 / 52                   | 42 / 42                |
+| Per-iteration/Fresh  | 51 / 52               | 47 / 50                   | 40 / 40                |
+
+The compiled policy evaluation shows performance characteristics that are generally comparable to engine evaluation, though with some notable differences. While single-threaded performance is very close between the systems, there are observable impacts from the compilation approach that become more apparent under different threading scenarios.
+
+**Key Observations:**
+- **Single-threaded performance**: Very close parity between systems, though results may vary between runs
+- **Threading behavior**: Engine evaluation demonstrates better scaling characteristics under higher thread contention (4+ threads)
+- **Multi-threaded impact**: Compiled policies show more pronounced performance degradation under thread contention in shared policy configurations
+- **Contention resistance**: Per-iteration compilation shows more consistent (though lower absolute) performance across thread counts
+- **Optimal usage**: Both systems achieve best results with minimal threading (1-4 threads), though engine evaluation maintains better performance at higher thread counts for shared configurations
+

benches/evaluation/compiled_policy_evaluation_benchmark.rs

@@ -0,0 +1,232 @@
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion, Throughput};
+use regorus::{compile_policy_with_entrypoint, CompiledPolicy, PolicyModule, Value};
+use std::collections::HashMap;
+use std::hint::black_box;
+use std::sync::{Arc, Barrier, Mutex};
+use std::thread;
+use std::time::Duration;
+
+mod policy_data;
+
+fn multi_threaded_compiled_eval(
+    num_threads: usize,
+    evals_per_thread: usize,
+    use_shared_policies: bool,
+    use_cloned_inputs: bool,
+) -> (std::time::Duration, HashMap<String, usize>, usize) {
+    // Complex policies with multiple valid inputs for each
+    let policies_with_inputs = policy_data::policies_with_inputs();
+
+    // Policy names for tracking
+    let policy_names = policy_data::policy_names()
+        .into_iter()
+        .map(|s| s.to_string())
+        .collect::<Vec<_>>();
+
+    // Pre-compile all policies and share them between threads (only if using shared policies)
+    let compiled_policies: Option<Arc<Vec<CompiledPolicy>>> = if use_shared_policies {
+        Some(Arc::new(
+            policies_with_inputs
+                .iter()
+                .map(|(policy, _)| {
+                    let module = PolicyModule {
+                        id: "policy.rego".into(),
+                        content: policy.as_str().into(),
+                    };
+                    compile_policy_with_entrypoint(
+                        Value::new_object(),
+                        &[module],
+                        "data.bench.allow".into(),
+                    )
+                    .unwrap()
+                })
+                .collect(),
+        ))
+    } else {
+        None
+    };
+
+    // Initialize policy evaluation counters
+    let policy_counters = Arc::new(Mutex::new(HashMap::new()));
+    for policy_name in &policy_names {
+        policy_counters
+            .lock()
+            .unwrap()
+            .insert(policy_name.to_string(), 0);
+    }
+    let total_evals = Arc::new(Mutex::new(0usize));
+
+    let barrier = Arc::new(Barrier::new(num_threads));
+    let mut handles = Vec::with_capacity(num_threads);
+
+    for thread_id in 0..num_threads {
+        let barrier = barrier.clone();
+        let policies_with_inputs = policies_with_inputs.clone();
+        let compiled_policies = compiled_policies.clone();
+        let policy_names = policy_names.clone();
+        let policy_counters = policy_counters.clone();
+        let total_evals = total_evals.clone();
+
+        handles.push(thread::spawn(move || {
+            let mut elapsed = std::time::Duration::ZERO;
+
+            // Pre-parse inputs if using cloned inputs
+            let parsed_inputs = if use_cloned_inputs {
+                Some(
+                    policies_with_inputs
+                        .iter()
+                        .map(|(_, inputs)| {
+                            inputs
+                                .iter()
+                                .map(|input_str| Value::from_json_str(input_str).unwrap())
+                                .collect::<Vec<_>>()
+                        })
+                        .collect::<Vec<_>>(),
+                )
+            } else {
+                None
+            };
+
+            barrier.wait();
+            for i in 0..evals_per_thread {
+                // Use different policy for each iteration - thread_id ensures different threads
+                // start with different policies for better load distribution
+                let policy_idx = (thread_id + i) % policies_with_inputs.len();
+                let (_, inputs) = &policies_with_inputs[policy_idx];
+
+                // Use different input for the same policy based on iteration - thread_id ensures
+                // different threads start with different inputs for better load distribution
+                let input_idx = (thread_id + i) % inputs.len();
+                let input = &inputs[input_idx];
+
+                let start = std::time::Instant::now();
+
+                let input_value = if use_cloned_inputs {
+                    parsed_inputs.as_ref().unwrap()[policy_idx][input_idx].clone()
+                } else {
+                    Value::from_json_str(input).unwrap()
+                };
+
+                let result = if let Some(ref compiled_policies_vec) = compiled_policies {
+                    // Use pre-compiled policy
+                    let compiled_policy = &compiled_policies_vec[policy_idx];
+                    compiled_policy.eval_with_input(input_value)
+                } else {
+                    // Compile policy in each iteration
+                    let (policy, _) = &policies_with_inputs[policy_idx];
+                    let module = PolicyModule {
+                        id: "policy.rego".into(),
+                        content: policy.as_str().into(),
+                    };
+                    let compiled_policy = compile_policy_with_entrypoint(
+                        Value::new_object(),
+                        &[module],
+                        "data.bench.allow".into(),
+                    )
+                    .unwrap();
+                    compiled_policy.eval_with_input(input_value)
+                };
+
+                elapsed += start.elapsed();
+
+                // Track total and successful evaluations
+                {
+                    let mut total = total_evals.lock().unwrap();
+                    *total += 1;
+                }
+                if result.is_ok() {
+                    if let Some(policy_name) = policy_names.get(policy_idx) {
+                        let mut counters = policy_counters.lock().unwrap();
+                        *counters.entry(policy_name.to_string()).or_insert(0) += 1;
+                    }
+                }
+            }
+            elapsed
+        }));
+    }
+
+    let mut total = std::time::Duration::ZERO;
+    for handle in handles {
+        total += handle.join().unwrap();
+    }
+
+    let final_counters = policy_counters.lock().unwrap().clone();
+    let total_evals = *total_evals.lock().unwrap();
+    (total, final_counters, total_evals)
+}
+
+fn criterion_benchmark(c: &mut Criterion) {
+    let max_threads = num_cpus::get() * 2;
+    println!(
+        "Running compiled policy benchmark with max_threads: {}",
+        max_threads
+    );
+
+    let evals_per_thread = 1000;
+
+    // Benchmark all combinations of compilation strategy and input strategy
+    for use_shared_policies in [true, false] {
+        for use_cloned_inputs in [true, false] {
+            let group_name = match (use_shared_policies, use_cloned_inputs) {
+                (true, true) => "compiled_shared_policies, cloned_inputs ",
+                (true, false) => "compiled_shared_policies, fresh_inputs  ",
+                (false, true) => "compiled_per_iteration , cloned_inputs ",
+                (false, false) => "compiled_per_iteration , fresh_inputs  ",
+            };
+
+            let mut group = c.benchmark_group(group_name);
+            group.measurement_time(Duration::from_secs(5));
+
+            // Test specific thread counts: powers of 2 + some intermediate values
+            let thread_counts: Vec<usize> = (1..=max_threads)
+                .filter(|&n| {
+                    n == 1 || // Always test single-threaded
+                    n % 2 == 0 || // Always test even threads
+                    n == max_threads // Maximum threads
+                })
+                .collect();
+
+            for threads in thread_counts {
+                let total_evals = threads * evals_per_thread;
+                group.throughput(Throughput::Elements(total_evals as u64));
+                group.bench_with_input(
+                    BenchmarkId::new("compiled_eval", format!(" {threads} threads")),
+                    &threads,
+                    |b, &threads| {
+                        b.iter_custom(|iters| {
+                            let evals_per_thread = evals_per_thread * (iters as usize);
+
+                            let (duration, policy_counters, total_evals_aggregated) = multi_threaded_compiled_eval(
+                                black_box(threads),
+                                black_box(evals_per_thread),
+                                black_box(use_shared_policies),
+                                black_box(use_cloned_inputs),
+                            );
+
+                            // Sanity check: Ensure the expected number of evaluations matches the actual number performed per iteration batch.
+                            // total_evals is the expected number for this batch, total_evals_aggregated is the sum over all iters.
+                            assert_eq!(total_evals, total_evals_aggregated/iters as usize);
+
+                            // On one iteration, print policy evaluation statistics
+                            if iters == 1 {
+                                // println!("\nCompiled Policy Evaluation Statistics:");
+                                for (policy_name, count) in &policy_counters {
+                                    // println!("  {}: {} evaluations", policy_name, count);
+                                    if *count == 0 {
+                                        println!("\x1b[31mERROR: Policy '{}' was never evaluated successfully!\x1b[0m", policy_name);
+                                    }
+                                }
+                            }
+
+                            duration
+                        });
+                    },
+                );
+            }
+            group.finish();
+        }
+    }
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

benches/evaluation/engine_evaluation_benchmark.md

@@ -0,0 +1,127 @@
+# Engine Evaluation Benchmark Results
+
+## Test Environment
+- **Platform**: Apple Silicon (M-Series)
+- **CPU**: 16 cores
+- **Architecture**: ARM64 (aarch64-apple-darwin)
+- **Rust Version**: 1.82.0
+- **Allocator**: mimalloc (default allocator)
+- **Benchmark Framework**: Criterion.rs
+- **Test Data**: 20,000 inputs per evaluation (1000 per thread)
+- **Policy**: Complex authorization policy with nested rules
+
+## Benchmark Overview
+
+The engine evaluation benchmark tests Regorus policy evaluation performance across multiple thread configurations (1-32 threads). It measures throughput (thousands of evaluations per second) for different combinations of engine and input data reuse strategies.
+
+## Configuration Combinations
+
+1. **Cloned Engines, Cloned Inputs**: Each thread uses its own engine and clones of parsed input data - optimal for performance
+2. **Cloned Engines, Fresh Inputs**: Each thread uses its own engine but parses new inputs each time
+3. **Fresh Engines, Cloned Inputs**: Each thread creates a new engine each iteration but reuses input data
+4. **Fresh Engines, Fresh Inputs**: Each thread creates new engines and parses new inputs for each iteration
+
+## Performance Results
+
+### Cloned Engines, Cloned Inputs (Best Performance)
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                       2.36 |                  423 |
+|       2 |                       4.85 |                  412 |
+|       4 |                       9.86 |                  406 |
+|       6 |                      15.02 |                  399 |
+|       8 |                      23.46 |                  341 |
+|      10 |                      33.34 |                  300 |
+|      12 |                      40.69 |                  295 |
+|      14 |                      48.26 |                  290 |
+|      16 |                      58.61 |                  273 |
+|      18 |                      77.35 |                  233 |
+|      20 |                      86.74 |                  231 |
+|      22 |                      94.17 |                  234 |
+|      24 |                     102.58 |                  234 |
+|      26 |                     110.17 |                  236 |
+|      28 |                     118.97 |                  235 |
+|      30 |                     126.54 |                  237 |
+|      32 |                     135.89 |                  235 |
+
+### Cloned Engines, Fresh Inputs
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                       3.24 |                  309 |
+|       2 |                       6.57 |                  304 |
+|       4 |                      13.47 |                  297 |
+|       6 |                      20.42 |                  294 |
+|       8 |                      30.01 |                  266 |
+|      10 |                      40.99 |                  244 |
+|      12 |                      49.99 |                  240 |
+|      14 |                      60.09 |                  233 |
+|      16 |                      73.95 |                  216 |
+|      18 |                      95.94 |                  188 |
+|      20 |                     105.24 |                  190 |
+|      22 |                     114.30 |                  192 |
+|      24 |                     124.67 |                  193 |
+|      26 |                     134.76 |                  193 |
+|      28 |                     145.16 |                  193 |
+|      30 |                     155.23 |                  193 |
+|      32 |                     165.42 |                  193 |
+
+### Fresh Engines, Cloned Inputs
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                      17.88 |                   56 |
+|       2 |                      36.32 |                   55 |
+|       4 |                      74.45 |                   54 |
+|       6 |                     112.95 |                   53 |
+|       8 |                     150.24 |                   53 |
+|      10 |                     189.61 |                   53 |
+|      12 |                     228.25 |                   53 |
+|      14 |                     297.37 |                   47 |
+|      16 |                     373.61 |                   43 |
+|      18 |                     426.46 |                   42 |
+|      20 |                     477.80 |                   42 |
+|      22 |                     523.00 |                   42 |
+|      24 |                     570.74 |                   42 |
+|      26 |                     619.92 |                   42 |
+|      28 |                     670.24 |                   42 |
+|      30 |                     717.47 |                   42 |
+|      32 |                     748.25 |                   43 |
+
+### Fresh Engines, Fresh Inputs
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                      18.69 |                   53 |
+|       2 |                      38.03 |                   53 |
+|       4 |                      77.82 |                   51 |
+|       6 |                     118.30 |                   51 |
+|       8 |                     157.65 |                   51 |
+|      10 |                     197.97 |                   51 |
+|      12 |                     239.05 |                   50 |
+|      14 |                     310.06 |                   45 |
+|      16 |                     391.36 |                   41 |
+|      18 |                     441.63 |                   41 |
+|      20 |                     495.88 |                   40 |
+|      22 |                     543.69 |                   40 |
+|      24 |                     591.51 |                   41 |
+|      26 |                     645.98 |                   40 |
+|      28 |                     697.37 |                   40 |
+|      30 |                     749.37 |                   40 |
+|      32 |                     784.63 |                   41 |
+
+## Analysis
+
+The benchmark results demonstrate the following performance characteristics with mimalloc as the default allocator:
+
+1. **Best Performance**: Cloned engines with cloned inputs consistently deliver the highest throughput
+2. **Configuration Performance Hierarchy**:
+   - Cloned engines, cloned inputs: Best performance (optimal configuration)
+   - Cloned engines, fresh inputs: ~27% reduction from optimal
+   - Fresh engines, cloned inputs: ~87% reduction from optimal
+   - Fresh engines, fresh inputs: ~87% reduction from optimal
+3. **Scaling Patterns with mimalloc**: 
+   - Performance degrades with increased thread count due to contention, but mimalloc provides better thread scaling characteristics
+   - Best throughput achieved at 1 thread for cloned engine configurations
+   - Fresh engine configurations show poor scaling across all thread counts
+   - The use of mimalloc as the default allocator has improved multi-threaded performance and reduced contention
+4. **Engine Creation Overhead**: Fresh engine creation is a significant performance bottleneck (~7-8x slower than cloned engines)
+5. **Input Processing**: Fresh input generation adds moderate overhead (~27% impact compared to cloned inputs)
+6. **Thread Contention**: Performance degradation occurs with higher thread counts across all configurations, though mimalloc helps mitigate some allocation-related contention

benches/evaluation/engine_evaluation_benchmark.rs

@@ -0,0 +1,228 @@
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion, Throughput};
+use regorus::{Engine, Value};
+use std::collections::HashMap;
+use std::hint::black_box;
+use std::sync::{Arc, Barrier, Mutex};
+use std::thread;
+use std::time::Duration;
+
+mod policy_data;
+
+fn multi_threaded_eval(
+    num_threads: usize,
+    evals_per_thread: usize,
+    use_cloned_engines: bool,
+    use_cloned_inputs: bool,
+) -> (std::time::Duration, HashMap<String, usize>, usize) {
+    // Complex policies with multiple valid inputs for each
+    let policies_with_inputs = policy_data::policies_with_inputs();
+
+    // Policy names for tracking
+    let policy_names = policy_data::policy_names()
+        .into_iter()
+        .map(|s| s.to_string())
+        .collect::<Vec<_>>();
+
+    // Initialize policy evaluation counters
+    let policy_counters = Arc::new(Mutex::new(HashMap::new()));
+    for policy_name in &policy_names {
+        policy_counters
+            .lock()
+            .unwrap()
+            .insert(policy_name.to_string(), 0);
+    }
+
+    let barrier = Arc::new(Barrier::new(num_threads));
+    let mut handles = Vec::with_capacity(num_threads);
+
+    let total_evals = Arc::new(Mutex::new(0usize));
+    for thread_id in 0..num_threads {
+        let barrier = barrier.clone();
+        let policies_with_inputs = policies_with_inputs.clone();
+        let policy_names = policy_names.clone();
+        let policy_counters = policy_counters.clone();
+        let total_evals = total_evals.clone();
+
+        handles.push(thread::spawn(move || {
+            let mut elapsed = std::time::Duration::ZERO;
+
+            // Pre-create engines if using cloned engines
+            let engines = if use_cloned_engines {
+                Some(
+                    policies_with_inputs
+                        .iter()
+                        .map(|(policy, _)| {
+                            let mut engine = Engine::new();
+                            engine
+                                .add_policy("policy.rego".to_string(), policy.to_string())
+                                .unwrap();
+                            {
+                                // Warm up the engine to ensure it's fully prepared for evaluation.
+                                // This prevents each cloned engine from repeating preparation work.
+                                engine.set_input(Value::new_object());
+                                let _ = engine.eval_rule("data.bench.allow".to_string());
+                            }
+                            engine
+                        })
+                        .collect::<Vec<_>>(),
+                )
+            } else {
+                None
+            };
+
+            // Pre-parse inputs if using cloned inputs
+            let parsed_inputs = if use_cloned_inputs {
+                Some(
+                    policies_with_inputs
+                        .iter()
+                        .map(|(_, inputs)| {
+                            inputs
+                                .iter()
+                                .map(|input_str| regorus::Value::from_json_str(input_str).unwrap())
+                                .collect::<Vec<_>>()
+                        })
+                        .collect::<Vec<_>>(),
+                )
+            } else {
+                None
+            };
+
+            barrier.wait();
+            for i in 0..evals_per_thread {
+                // Use different policy for each iteration - thread_id ensures different threads
+                // start with different policies for better load distribution
+                let policy_idx = (thread_id + i) % policies_with_inputs.len();
+                let (policy, inputs) = &policies_with_inputs[policy_idx];
+
+                // Use different input for the same policy based on iteration - thread_id ensures
+                // different threads start with different inputs for better load distribution
+                let input_idx = (thread_id + i) % inputs.len();
+                let input = &inputs[input_idx];
+
+                let start = std::time::Instant::now();
+
+                let result = {
+                    let mut engine = if use_cloned_engines {
+                        engines.as_ref().unwrap()[policy_idx].clone()
+                    } else {
+                        let mut engine = Engine::new();
+                        engine
+                            .add_policy("policy.rego".to_string(), policy.to_string())
+                            .unwrap();
+                        engine
+                    };
+
+                    let input_value = if use_cloned_inputs {
+                        parsed_inputs.as_ref().unwrap()[policy_idx][input_idx].clone()
+                    } else {
+                        regorus::Value::from_json_str(input).unwrap()
+                    };
+
+                    engine.set_input(input_value);
+
+                    engine.eval_rule("data.bench.allow".to_string())
+
+                    // Engine cleanup/drop time is included in measurement to reflect
+                    // real-world total cost of policy evaluation lifecycle
+                };
+                elapsed += start.elapsed();
+
+                // Track total and successful evaluations
+                {
+                    let mut total = total_evals.lock().unwrap();
+                    *total += 1;
+                }
+                if result.is_ok() {
+                    if let Some(policy_name) = policy_names.get(policy_idx) {
+                        let mut counters = policy_counters.lock().unwrap();
+                        *counters.entry(policy_name.to_string()).or_insert(0) += 1;
+                    }
+                }
+            }
+            elapsed
+        }));
+    }
+
+    let mut total = std::time::Duration::ZERO;
+    for handle in handles {
+        total += handle.join().unwrap();
+    }
+
+    let final_counters = policy_counters.lock().unwrap().clone();
+    let total_evals = *total_evals.lock().unwrap();
+    (total, final_counters, total_evals)
+}
+
+fn criterion_benchmark(c: &mut Criterion) {
+    let max_threads = num_cpus::get() * 2;
+    println!("Running benchmark with max_threads: {}", max_threads);
+
+    let evals_per_thread = 1000;
+
+    // Benchmark all combinations of cloned engines and inputs
+    for use_cloned_engines in [true, false] {
+        for use_cloned_inputs in [true, false] {
+            let group_name = match (use_cloned_engines, use_cloned_inputs) {
+                (true, true) => "cloned_engines , cloned_inputs ",
+                (true, false) => "cloned_engines , fresh_inputs  ",
+                (false, true) => "fresh_engines  , cloned_inputs ",
+                (false, false) => "fresh_engines  , fresh_inputs  ",
+            };
+
+            let mut group = c.benchmark_group(group_name);
+            group.measurement_time(Duration::from_secs(5));
+
+            // Test specific thread counts: powers of 2 + some intermediate values
+            let thread_counts: Vec<usize> = (1..=max_threads)
+                .filter(|&n| {
+                    n == 1 || // Always test single-threaded
+                    n % 2 == 0 || // Always test even threads
+                    n == max_threads // Maximum threads
+                })
+                .collect();
+
+            for threads in thread_counts {
+                let total_evals = threads * evals_per_thread;
+                group.throughput(Throughput::Elements(total_evals as u64));
+                group.bench_with_input(
+                    BenchmarkId::new("eval", format!(" {threads} threads")),
+                    &threads,
+                    |b, &threads| {
+                        b.iter_custom(|iters| {
+                            let evals_per_thread = evals_per_thread * (iters as usize);
+
+                            let (duration, policy_counters, total_evals_aggregated) = multi_threaded_eval(
+                                black_box(threads),
+                                black_box(evals_per_thread),
+                                black_box(use_cloned_engines),
+                                black_box(use_cloned_inputs),
+                            );
+
+
+                            // Sanity check: Ensure the expected number of evaluations matches the actual number performed per iteration batch.
+                            // total_evals is the expected number for this batch, total_evals_aggregated is the sum over all iters.
+                            assert_eq!(total_evals, total_evals_aggregated/iters as usize);
+
+                            // On one iteration, print policy evaluation statistics
+                            if iters == 1 {
+                                // println!("\nPolicy Evaluation Statistics:");
+                                for (policy_name, count) in &policy_counters {
+                                    // println!("  {}: {} evaluations", policy_name, count);
+                                    if *count == 0 {
+                                        println!("\x1b[31mERROR: Policy '{}' was never evaluated successfully!\x1b[0m", policy_name);
+                                    }
+                                }
+                            }
+
+                            duration
+                        });
+                    },
+                );
+            }
+            group.finish();
+        }
+    }
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

benches/evaluation/policy_data.rs

@@ -0,0 +1,117 @@
+// This module provides the full set of policies, inputs, and policy names for evaluation benchmarks.
+// Policies and inputs are now loaded from external files.
+
+use std::fs;
+use std::path::Path;
+
+pub fn policies_with_inputs() -> Vec<(String, Vec<String>)> {
+    let policy_with_input_files = [
+        (
+            "rbac_policy.rego",
+            vec!["rbac_input.json", "rbac_input2.json", "rbac_input3.json"],
+        ),
+        (
+            "api_access_policy.rego",
+            vec![
+                "api_access_input.json",
+                "api_access_input2.json",
+                "api_access_input3.json",
+            ],
+        ),
+        (
+            "data_sensitivity_policy.rego",
+            vec![
+                "data_sensitivity_input.json",
+                "data_sensitivity_input2.json",
+                "data_sensitivity_input3.json",
+            ],
+        ),
+        (
+            "time_based_policy.rego",
+            vec![
+                "time_based_input.json",
+                "time_based_input2.json",
+                "time_based_input3.json",
+            ],
+        ),
+        (
+            "data_processing_policy.rego",
+            vec![
+                "data_processing_input.json",
+                "data_processing_input2.json",
+                "data_processing_input3.json",
+            ],
+        ),
+        (
+            "azure_vm_policy.rego",
+            vec![
+                "azure_vm_input.json",
+                "azure_vm_input2.json",
+                "azure_vm_input3.json",
+            ],
+        ),
+        (
+            "azure_storage_policy.rego",
+            vec![
+                "azure_storage_input.json",
+                "azure_storage_input2.json",
+                "azure_storage_input3.json",
+            ],
+        ),
+        (
+            "azure_keyvault_policy.rego",
+            vec![
+                "azure_keyvault_input.json",
+                "azure_keyvault_input2.json",
+                "azure_keyvault_input3.json",
+            ],
+        ),
+        (
+            "azure_nsg_policy.rego",
+            vec![
+                "azure_nsg_input.json",
+                "azure_nsg_input2.json",
+                "azure_nsg_input3.json",
+            ],
+        ),
+    ];
+
+    let mut policies_and_inputs = Vec::new();
+    let base_dir = Path::new(env!("CARGO_MANIFEST_DIR"))
+        .join("benches")
+        .join("evaluation")
+        .join("test_data");
+
+    for (policy_file, input_files) in policy_with_input_files.iter() {
+        let policy_path = base_dir.join("policies").join(policy_file);
+
+        let policy_content = fs::read_to_string(&policy_path)
+            .unwrap_or_else(|e| panic!("Failed to read policy file {:?}: {}", policy_path, e));
+
+        let mut input_contents = Vec::new();
+        for input_file in input_files {
+            let input_path = base_dir.join("inputs").join(input_file);
+            let input_content = fs::read_to_string(&input_path)
+                .unwrap_or_else(|e| panic!("Failed to read input file {:?}: {}", input_path, e));
+            input_contents.push(input_content);
+        }
+
+        policies_and_inputs.push((policy_content, input_contents));
+    }
+
+    policies_and_inputs
+}
+
+pub fn policy_names() -> Vec<&'static str> {
+    vec![
+        "rbac_policy",
+        "api_access_policy",
+        "data_sensitivity_policy",
+        "time_based_policy",
+        "data_processing_policy",
+        "azure_vm_policy",
+        "azure_storage_policy",
+        "azure_keyvault_policy",
+        "azure_nsg_policy",
+    ]
+}

benches/evaluation/test_data/inputs/api_access_input.json

@@ -0,0 +1,15 @@
+{
+  "request": {
+    "method": "GET",
+    "path": "/api/v1/users/123"
+  },
+  "user": {
+    "id": "user123",
+    "scope": ["read:users", "write:users"],
+    "department": "engineering"
+  },
+  "resource": {
+    "owner": "user123",
+    "type": "user"
+  }
+}

benches/evaluation/test_data/inputs/api_access_input2.json

@@ -0,0 +1,15 @@
+{
+  "request": {
+    "method": "POST",
+    "path": "/api/v1/users"
+  },
+  "user": {
+    "id": "user456",
+    "scope": ["write:users", "admin:users"],
+    "department": "engineering"
+  },
+  "resource": {
+    "owner": "user456",
+    "type": "user"
+  }
+}

benches/evaluation/test_data/inputs/api_access_input3.json

@@ -0,0 +1,15 @@
+{
+  "request": {
+    "method": "DELETE",
+    "path": "/api/v1/users/789"
+  },
+  "user": {
+    "id": "admin123",
+    "scope": ["admin:users"],
+    "department": "security"
+  },
+  "resource": {
+    "owner": "user789",
+    "type": "user"
+  }
+}

benches/evaluation/test_data/inputs/azure_keyvault_input.json

@@ -0,0 +1,16 @@
+{
+  "vault": {
+    "name": "mykeyvault",
+    "location": "eastus",
+    "enableSoftDelete": true,
+    "softDeleteRetentionInDays": 90,
+    "enablePurgeProtection": true,
+    "networkAcls": {
+      "defaultAction": "Deny",
+      "bypass": "AzureServices"
+    },
+    "tags": {
+      "environment": "production"
+    }
+  }
+}

benches/evaluation/test_data/inputs/azure_keyvault_input2.json

@@ -0,0 +1,16 @@
+{
+  "vault": {
+    "name": "devkeyvault",
+    "location": "westus2",
+    "enableSoftDelete": true,
+    "softDeleteRetentionInDays": 30,
+    "enablePurgeProtection": false,
+    "networkAcls": {
+      "defaultAction": "Allow",
+      "bypass": "AzureServices"
+    },
+    "tags": {
+      "environment": "development"
+    }
+  }
+}

benches/evaluation/test_data/inputs/azure_keyvault_input3.json

@@ -0,0 +1,16 @@
+{
+  "vault": {
+    "name": "prodkeyvault",
+    "location": "eastus",
+    "enableSoftDelete": true,
+    "softDeleteRetentionInDays": 90,
+    "enablePurgeProtection": true,
+    "networkAcls": {
+      "defaultAction": "Deny",
+      "bypass": "AzureServices"
+    },
+    "tags": {
+      "environment": "production"
+    }
+  }
+}

benches/evaluation/test_data/inputs/azure_nsg_input.json

@@ -0,0 +1,13 @@
+{
+  "operation": "Microsoft.Network/networkSecurityGroups/securityRules/write",
+  "rule": {
+    "direction": "Inbound",
+    "access": "Allow",
+    "protocol": "TCP",
+    "sourceAddressPrefix": "10.0.0.0/24",
+    "sourcePortRange": "*",
+    "destinationAddressPrefix": "*",
+    "destinationPortRange": "80",
+    "priority": 1001
+  }
+}

benches/evaluation/test_data/inputs/azure_nsg_input2.json

@@ -0,0 +1,13 @@
+{
+  "operation": "Microsoft.Network/networkSecurityGroups/securityRules/write",
+  "rule": {
+    "direction": "Inbound",
+    "access": "Allow",
+    "protocol": "TCP",
+    "sourceAddressPrefix": "172.16.0.0/16",
+    "sourcePortRange": "*",
+    "destinationAddressPrefix": "*",
+    "destinationPortRange": "22",
+    "priority": 1200
+  }
+}

benches/evaluation/test_data/inputs/azure_nsg_input3.json

@@ -0,0 +1,13 @@
+{
+  "operation": "Microsoft.Network/networkSecurityGroups/securityRules/write",
+  "rule": {
+    "direction": "Inbound",
+    "access": "Allow",
+    "protocol": "TCP",
+    "sourceAddressPrefix": "203.0.113.0/24",
+    "sourcePortRange": "*",
+    "destinationAddressPrefix": "*",
+    "destinationPortRange": "443",
+    "priority": 300
+  }
+}

benches/evaluation/test_data/inputs/azure_storage_input.json

@@ -0,0 +1,15 @@
+{
+  "account": {
+    "name": "mystorageaccount",
+    "tier": "Standard",
+    "replication": "LRS",
+    "location": "eastus",
+    "tags": {
+      "environment": "production"
+    }
+  },
+  "container": {
+    "name": "data",
+    "publicAccess": "None"
+  }
+}

benches/evaluation/test_data/inputs/azure_storage_input2.json

@@ -0,0 +1,15 @@
+{
+  "account": {
+    "name": "devstorageaccount",
+    "tier": "Premium",
+    "replication": "LRS",
+    "location": "westus2",
+    "tags": {
+      "environment": "production"
+    }
+  },
+  "container": {
+    "name": "logs",
+    "publicAccess": "None"
+  }
+}

benches/evaluation/test_data/inputs/azure_storage_input3.json

@@ -0,0 +1,15 @@
+{
+  "account": {
+    "name": "prodstorageaccount",
+    "tier": "Standard",
+    "replication": "GRS",
+    "location": "eastus",
+    "tags": {
+      "environment": "production"
+    }
+  },
+  "container": {
+    "name": "backups",
+    "publicAccess": "None"
+  }
+}

benches/evaluation/test_data/inputs/azure_vm_input.json

@@ -0,0 +1,14 @@
+{
+  "vm": {
+    "size": "Standard_D2s_v3",
+    "os": "Linux",
+    "location": "eastus",
+    "tags": {
+      "environment": "production",
+      "department": "engineering"
+    }
+  },
+  "user": {
+    "department": "engineering"
+  }
+}

benches/evaluation/test_data/inputs/azure_vm_input2.json

@@ -0,0 +1,14 @@
+{
+  "vm": {
+    "size": "Standard_B1s",
+    "os": "Windows",
+    "location": "westus2",
+    "tags": {
+      "environment": "dev",
+      "department": "marketing"
+    }
+  },
+  "user": {
+    "department": "marketing"
+  }
+}

benches/evaluation/test_data/inputs/azure_vm_input3.json

@@ -0,0 +1,14 @@
+{
+  "vm": {
+    "size": "Standard_D4s_v3",
+    "os": "Linux",
+    "location": "eastus",
+    "tags": {
+      "environment": "production",
+      "department": "engineering"
+    }
+  },
+  "user": {
+    "department": "engineering"
+  }
+}

benches/evaluation/test_data/inputs/data_processing_input.json

@@ -0,0 +1,16 @@
+{
+  "operation": "collect",
+  "data": {
+    "type": "email",
+    "source": "user_input"
+  },
+  "consent": {
+    "given": true,
+    "purpose": "marketing",
+    "date": "2023-01-15"
+  },
+  "user": {
+    "age": 25,
+    "location": "US"
+  }
+}

benches/evaluation/test_data/inputs/data_processing_input2.json

@@ -0,0 +1,16 @@
+{
+  "operation": "process",
+  "data": {
+    "type": "survey_response",
+    "source": "user_input"
+  },
+  "consent": {
+    "given": true,
+    "purpose": "analytics",
+    "date": "2023-06-15"
+  },
+  "user": {
+    "age": 30,
+    "location": "US"
+  }
+}

benches/evaluation/test_data/inputs/data_processing_input3.json

@@ -0,0 +1,16 @@
+{
+  "operation": "delete",
+  "data": {
+    "type": "user_profile",
+    "source": "database"
+  },
+  "consent": {
+    "given": false,
+    "purpose": "none",
+    "date": "2022-01-01"
+  },
+  "user": {
+    "age": 16,
+    "location": "EU"
+  }
+}

benches/evaluation/test_data/inputs/data_sensitivity_input.json

@@ -0,0 +1,13 @@
+{
+  "data": {
+    "type": "user_profile",
+    "classification": "personal",
+    "contains_pii": true,
+    "region": "EU"
+  },
+  "user": {
+    "clearance": "confidential",
+    "location": "EU"
+  },
+  "operation": "read"
+}

benches/evaluation/test_data/inputs/data_sensitivity_input2.json

@@ -0,0 +1,13 @@
+{
+  "data": {
+    "type": "financial_report",
+    "classification": "confidential",
+    "contains_pii": false,
+    "region": "US"
+  },
+  "user": {
+    "clearance": "secret",
+    "location": "US"
+  },
+  "operation": "read"
+}

benches/evaluation/test_data/inputs/data_sensitivity_input3.json

@@ -0,0 +1,13 @@
+{
+  "data": {
+    "type": "public_announcement",
+    "classification": "public",
+    "contains_pii": false,
+    "region": "GLOBAL"
+  },
+  "user": {
+    "clearance": "public",
+    "location": "EU"
+  },
+  "operation": "read"
+}

benches/evaluation/test_data/inputs/rbac_input.json

@@ -0,0 +1,12 @@
+{
+  "user": {
+    "name": "alice",
+    "roles": ["viewer", "editor"]
+  },
+  "resource": {
+    "name": "document1",
+    "type": "document",
+    "owner": "alice"
+  },
+  "action": "read"
+}

benches/evaluation/test_data/inputs/rbac_input2.json

@@ -0,0 +1,12 @@
+{
+  "user": {
+    "name": "bob",
+    "roles": ["admin"]
+  },
+  "resource": {
+    "name": "document2",
+    "type": "document",
+    "owner": "bob"
+  },
+  "action": "write"
+}

benches/evaluation/test_data/inputs/rbac_input3.json

@@ -0,0 +1,12 @@
+{
+  "user": {
+    "name": "charlie",
+    "roles": ["viewer"]
+  },
+  "resource": {
+    "name": "document3",
+    "type": "document",
+    "owner": "alice"
+  },
+  "action": "read"
+}

benches/evaluation/test_data/inputs/time_based_input.json

@@ -0,0 +1,11 @@
+{
+  "time": "09:30:00",
+  "day": "monday",
+  "user": {
+    "role": "employee",
+    "shift": "day"
+  },
+  "request": {
+    "urgent": false
+  }
+}

benches/evaluation/test_data/inputs/time_based_input2.json

@@ -0,0 +1,11 @@
+{
+  "time": "14:30:00",
+  "day": "wednesday",
+  "user": {
+    "role": "employee",
+    "shift": "day"
+  },
+  "request": {
+    "urgent": false
+  }
+}

benches/evaluation/test_data/inputs/time_based_input3.json

@@ -0,0 +1,11 @@
+{
+  "time": "22:00:00",
+  "day": "friday",
+  "user": {
+    "role": "admin",
+    "shift": "night"
+  },
+  "request": {
+    "urgent": true
+  }
+}

benches/evaluation/test_data/policies/api_access_policy.rego

@@ -0,0 +1,13 @@
+package bench
+
+default allow := false
+
+valid_api_paths := ["/api/v1/", "/api/v2/", "/api/v3/"]
+
+allow if {
+    input.request.method == "GET"
+    some path in valid_api_paths
+    startswith(input.request.path, path)
+    input.user.authenticated == true
+    time.now_ns() - input.user.login_time < 86400000000000  # 24 hours in nanoseconds
+}

benches/evaluation/test_data/policies/azure_keyvault_policy.rego

@@ -0,0 +1,28 @@
+package bench
+
+default allow := false
+
+# Azure Key Vault access policy
+valid_operations := [
+    "Microsoft.KeyVault/vaults/keys/read",
+    "Microsoft.KeyVault/vaults/secrets/read",
+    "Microsoft.KeyVault/vaults/certificates/read"
+]
+
+vault_admins := ["admin@company.com", "security@company.com"]
+
+allow if {
+    input.operation in valid_operations
+    input.principal.type == "ServicePrincipal"
+    input.principal.appId != ""
+    input.resource.properties.enableSoftDelete == true
+    input.resource.properties.enablePurgeProtection == true
+    time.now_ns() - input.principal.createdTime < 31536000000000000  # Less than 1 year old
+}
+
+allow if {
+    input.operation in valid_operations
+    input.principal.type == "User"
+    input.principal.userPrincipalName in vault_admins
+    input.context.conditionalAccess.compliant == true
+}

benches/evaluation/test_data/policies/azure_nsg_policy.rego

@@ -0,0 +1,31 @@
+package bench
+
+default allow := false
+
+# Azure Network Security Group rules policy
+dangerous_ports := [22, 3389, 1433, 3306, 5432, 6379, 27017]
+internal_networks := ["10.0.0.0/8", "172.16.0.0/12", "192.168.0.0/16"]
+
+is_internal_source if {
+    some network in internal_networks
+    net.cidr_contains(network, input.rule.sourceAddressPrefix)
+}
+
+allow if {
+    input.operation == "Microsoft.Network/networkSecurityGroups/securityRules/write"
+    input.rule.direction == "Inbound"
+    input.rule.access == "Allow"
+    input.rule.destinationPortRange != "*"
+    not input.rule.destinationPortRange in dangerous_ports
+    input.rule.sourceAddressPrefix != "*"
+    input.rule.sourceAddressPrefix != "Internet"
+}
+
+allow if {
+    input.operation == "Microsoft.Network/networkSecurityGroups/securityRules/write"
+    input.rule.direction == "Inbound"
+    input.rule.access == "Allow"
+    input.rule.destinationPortRange in dangerous_ports
+    is_internal_source
+    input.rule.priority >= 1000
+}

benches/evaluation/test_data/policies/azure_storage_policy.rego

@@ -0,0 +1,17 @@
+package bench
+
+default allow := false
+
+# Azure Storage Account security policy
+required_encryption_algorithms := ["AES256", "RSA-OAEP"]
+
+allow if {
+    input.operation == "Microsoft.Storage/storageAccounts/write"
+    input.resource.properties.supportsHttpsTrafficOnly == true
+    input.resource.properties.minimumTlsVersion == "TLS1_2"
+    input.resource.properties.encryption.services.blob.enabled == true
+    input.resource.properties.encryption.keySource == "Microsoft.Storage"
+    input.resource.properties.allowBlobPublicAccess == false
+    input.resource.properties.networkAcls.defaultAction == "Deny"
+    count(input.resource.properties.networkAcls.ipRules) > 0
+}

benches/evaluation/test_data/policies/azure_vm_policy.rego

@@ -0,0 +1,20 @@
+package bench
+
+default allow := false
+
+# Azure VM deployment policy
+allowed_vm_sizes := [
+    "Standard_B1s", "Standard_B2s", "Standard_B4ms",
+    "Standard_D2s_v3", "Standard_D4s_v3", "Standard_F2s_v2"
+]
+
+allowed_regions := ["eastus", "westus2", "northeurope", "southeastasia"]
+
+allow if {
+    input.operation == "Microsoft.Compute/virtualMachines/write"
+    input.resource.properties.hardwareProfile.vmSize in allowed_vm_sizes
+    input.resource.location in allowed_regions
+    input.resource.properties.osProfile.adminPassword == null  # Require SSH keys
+    count(input.resource.tags) > 0  # Must have tags
+    input.resource.tags.environment in ["dev", "test", "prod"]
+}

benches/evaluation/test_data/policies/data_processing_policy.rego

@@ -0,0 +1,28 @@
+package bench
+
+default allow := false
+
+# Complex data filtering and aggregation
+sensitive_fields := ["ssn", "credit_card", "password"]
+
+contains_sensitive_data if {
+    some field in sensitive_fields
+    object.get(input.data, field, null) != null
+}
+
+user_clearance_level := object.get(input.user.attributes, "clearance", 0)
+
+required_clearance := 3 if contains_sensitive_data else := 1
+
+allow if {
+    user_clearance_level >= required_clearance
+    input.operation in ["read", "export"]
+    count(input.data) > 0
+    count(input.data) <= 1000  # Limit data size
+}
+
+allow if {
+    input.user.role == "data_processor"
+    input.operation == "transform"
+    not contains_sensitive_data
+}

benches/evaluation/test_data/policies/data_sensitivity_policy.rego

@@ -0,0 +1,25 @@
+package bench
+
+default allow := false
+
+rbac_roles := {
+    "admin": ["read", "write", "delete", "admin"],
+    "manager": ["read", "write"],
+    "user": ["read"]
+}
+
+user_permissions contains perm if {
+    some role in input.user.roles
+    perm := rbac_roles[role][_]
+}
+
+allow if {
+    input.action in user_permissions
+    input.resource.owner == input.user.id
+}
+
+allow if {
+    input.action in user_permissions
+    input.resource.public == true
+    input.action == "read"
+}

benches/evaluation/test_data/policies/rbac_policy.rego

@@ -0,0 +1,10 @@
+package bench
+
+default allow := false
+
+allow if {
+    input.user.role == "admin"
+    input.action in ["read", "write", "delete"]
+    input.resource.classification in ["public", "internal"]
+    count(input.user.permissions) > 0
+}

benches/evaluation/test_data/policies/time_based_policy.rego

@@ -0,0 +1,23 @@
+package bench
+
+default allow := false
+
+# Time-based access control with complex conditions
+business_hours if {
+    hour := time.clock([time.now_ns(), "America/New_York"])[0]
+    hour >= 9
+    hour < 17
+}
+
+allow if {
+    input.user.department in ["engineering", "product"]
+    input.action == "deploy"
+    business_hours
+    count([x | x := input.approvals[_]; x.status == "approved"]) >= 2
+}
+
+allow if {
+    input.user.emergency_access == true
+    input.action in ["read", "diagnose"]
+    input.justification != ""
+}

benches/normalization_benchmark.rs

@@ -0,0 +1,560 @@
+use std::hint::black_box;
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion};
+use regorus::languages::azure_policy::aliases::{denormalizer, normalizer, AliasRegistry};
+use regorus::Value;
+use serde_json::json;
+
+// ─── Alias catalog (reused across benchmarks) ───────────────────────────────
+
+const ALIASES_JSON: &str = r#"[
+  {
+    "namespace": "Microsoft.Network",
+    "resourceTypes": [
+      {
+        "resourceType": "networkSecurityGroups",
+        "aliases": [
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].protocol",
+            "defaultPath": "properties.securityRules[*].properties.protocol",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].access",
+            "defaultPath": "properties.securityRules[*].properties.access",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].priority",
+            "defaultPath": "properties.securityRules[*].properties.priority",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].direction",
+            "defaultPath": "properties.securityRules[*].properties.direction",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].sourceAddressPrefix",
+            "defaultPath": "properties.securityRules[*].properties.sourceAddressPrefix",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].destinationPortRange",
+            "defaultPath": "properties.securityRules[*].properties.destinationPortRange",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].name",
+            "defaultPath": "properties.securityRules[*].name",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/defaultSecurityRules[*].protocol",
+            "defaultPath": "properties.defaultSecurityRules[*].properties.protocol",
+            "paths": []
+          }
+        ]
+      }
+    ]
+  },
+  {
+    "namespace": "Microsoft.Storage",
+    "resourceTypes": [
+      {
+        "resourceType": "storageAccounts",
+        "aliases": [
+          {
+            "name": "Microsoft.Storage/storageAccounts/supportsHttpsTrafficOnly",
+            "defaultPath": "properties.supportsHttpsTrafficOnly",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Storage/storageAccounts/accessTier",
+            "defaultPath": "properties.accessTier",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Storage/storageAccounts/isHnsEnabled",
+            "defaultPath": "properties.isHnsEnabled",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Storage/storageAccounts/minimumTlsVersion",
+            "defaultPath": "properties.minimumTlsVersion",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Storage/storageAccounts/allowBlobPublicAccess",
+            "defaultPath": "properties.allowBlobPublicAccess",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Storage/storageAccounts/sku.name",
+            "defaultPath": "sku.name",
+            "paths": []
+          }
+        ]
+      }
+    ]
+  }
+]"#;
+
+fn build_registry() -> AliasRegistry {
+    let mut reg = AliasRegistry::new();
+    reg.load_from_json(ALIASES_JSON).unwrap();
+    reg
+}
+
+/// Convert a serde_json::Value to regorus::Value.
+fn to_regorus(v: serde_json::Value) -> Value {
+    Value::from(v)
+}
+
+// ─── Input resources ────────────────────────────────────────────────────────
+
+fn simple_storage_resource() -> Value {
+    to_regorus(json!({
+        "name": "myStorageAccount",
+        "type": "Microsoft.Storage/storageAccounts",
+        "location": "westus2",
+        "kind": "StorageV2",
+        "sku": { "name": "Standard_LRS", "tier": "Standard" },
+        "tags": { "environment": "production", "team": "platform" },
+        "properties": {
+            "supportsHttpsTrafficOnly": true,
+            "accessTier": "Hot",
+            "isHnsEnabled": false,
+            "minimumTlsVersion": "TLS1_2",
+            "allowBlobPublicAccess": false
+        }
+    }))
+}
+
+fn nsg_resource(rule_count: usize) -> Value {
+    let rules: Vec<serde_json::Value> = (0..rule_count)
+        .map(|i| {
+            json!({
+                "name": format!("rule-{}", i),
+                "properties": {
+                    "protocol": "Tcp",
+                    "access": if i % 2 == 0 { "Allow" } else { "Deny" },
+                    "priority": 100 + i,
+                    "direction": "Inbound",
+                    "sourceAddressPrefix": format!("10.0.{}.0/24", i % 256),
+                    "destinationPortRange": format!("{}", 80 + i)
+                }
+            })
+        })
+        .collect();
+
+    to_regorus(json!({
+        "name": "myNsg",
+        "type": "Microsoft.Network/networkSecurityGroups",
+        "location": "eastus",
+        "properties": {
+            "securityRules": rules
+        }
+    }))
+}
+
+// ─── Benchmarks ─────────────────────────────────────────────────────────────
+
+fn bench_normalize_simple(c: &mut Criterion) {
+    let registry = build_registry();
+    let resource = simple_storage_resource();
+
+    c.bench_function("normalize/simple_storage", |b| {
+        b.iter(|| normalizer::normalize(black_box(&resource), Some(&registry), None))
+    });
+}
+
+fn bench_normalize_no_aliases(c: &mut Criterion) {
+    let resource = simple_storage_resource();
+
+    c.bench_function("normalize/simple_no_aliases", |b| {
+        b.iter(|| normalizer::normalize(black_box(&resource), None, None))
+    });
+}
+
+fn bench_normalize_nsg_scaling(c: &mut Criterion) {
+    let registry = build_registry();
+    let mut group = c.benchmark_group("normalize/nsg_rules");
+
+    for rule_count in [5, 20, 100] {
+        let resource = nsg_resource(rule_count);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(rule_count),
+            &resource,
+            |b, res| b.iter(|| normalizer::normalize(black_box(res), Some(&registry), None)),
+        );
+    }
+    group.finish();
+}
+
+fn bench_denormalize_simple(c: &mut Criterion) {
+    let registry = build_registry();
+    let resource = simple_storage_resource();
+    let normalized = normalizer::normalize(&resource, Some(&registry), None);
+
+    c.bench_function("denormalize/simple_storage", |b| {
+        b.iter(|| denormalizer::denormalize(black_box(&normalized), Some(&registry), None))
+    });
+}
+
+fn bench_denormalize_nsg_scaling(c: &mut Criterion) {
+    let registry = build_registry();
+    let mut group = c.benchmark_group("denormalize/nsg_rules");
+
+    for rule_count in [5, 20, 100] {
+        let resource = nsg_resource(rule_count);
+        let normalized = normalizer::normalize(&resource, Some(&registry), None);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(rule_count),
+            &normalized,
+            |b, norm| b.iter(|| denormalizer::denormalize(black_box(norm), Some(&registry), None)),
+        );
+    }
+    group.finish();
+}
+
+fn bench_round_trip(c: &mut Criterion) {
+    let registry = build_registry();
+    let resource = nsg_resource(20);
+
+    c.bench_function("round_trip/nsg_20_rules", |b| {
+        b.iter(|| {
+            let n = normalizer::normalize(black_box(&resource), Some(&registry), None);
+            denormalizer::denormalize(&n, Some(&registry), None)
+        })
+    });
+}
+
+fn bench_normalize_and_wrap(c: &mut Criterion) {
+    let registry = build_registry();
+    let resource = nsg_resource(20);
+    let context = to_regorus(json!({"resourceGroup": {"name": "rg1"}}));
+    let parameters = to_regorus(json!({"env": "prod"}));
+
+    c.bench_function("normalize_and_wrap/nsg_20_rules", |b| {
+        b.iter(|| {
+            registry.normalize_and_wrap(
+                black_box(&resource),
+                None,
+                Some(context.clone()),
+                Some(parameters.clone()),
+            )
+        })
+    });
+}
+
+fn bench_registry_load(c: &mut Criterion) {
+    c.bench_function("registry/load_from_json", |b| {
+        b.iter(|| {
+            let mut reg = AliasRegistry::new();
+            reg.load_from_json(black_box(ALIASES_JSON)).unwrap();
+            reg
+        })
+    });
+}
+
+// ─── Large-payload benchmarks ───────────────────────────────────────────────
+//
+// These stress the hot paths identified in the performance analysis:
+// - Nested set helpers (alias-heavy catalog with deep properties)
+// - Array element remap/cleanup/rewrap (large sub-resource arrays)
+// - Scalar denormalization lookups (many aliases × many fields)
+
+/// Build a large alias catalog with `n` scalar aliases for storage accounts.
+/// Each alias maps to a nested `properties.section_i.field_j` path, creating
+/// deep nested-set workloads.
+fn large_alias_catalog(n: usize) -> String {
+    let mut aliases = Vec::new();
+    for i in 0..n {
+        let section = i / 10;
+        let field = i % 10;
+        aliases.push(format!(
+            r#"{{
+                "name": "Microsoft.Storage/storageAccounts/section{section}Field{field}",
+                "defaultPath": "properties.section{section}.field{field}",
+                "paths": []
+            }}"#,
+        ));
+    }
+    format!(
+        r#"[{{
+            "namespace": "Microsoft.Storage",
+            "resourceTypes": [{{
+                "resourceType": "storageAccounts",
+                "aliases": [{aliases}]
+            }}]
+        }}]"#,
+        aliases = aliases.join(",")
+    )
+}
+
+/// Build a storage account resource whose `properties` contain nested sections
+/// matching the large alias catalog.
+fn large_storage_resource(alias_count: usize) -> Value {
+    let mut sections = serde_json::Map::new();
+    for i in 0..alias_count {
+        let section = i / 10;
+        let field = i % 10;
+        let section_key = format!("section{section}");
+        let section_obj = sections
+            .entry(section_key)
+            .or_insert_with(|| serde_json::Value::Object(serde_json::Map::new()));
+        if let serde_json::Value::Object(m) = section_obj {
+            m.insert(format!("field{field}"), serde_json::Value::from(i));
+        }
+    }
+    Value::from(json!({
+        "name": "bigStorage",
+        "type": "Microsoft.Storage/storageAccounts",
+        "location": "westus2",
+        "properties": sections
+    }))
+}
+
+fn bench_normalize_large_catalog(c: &mut Criterion) {
+    let mut group = c.benchmark_group("normalize/large_catalog");
+    for alias_count in [50, 200] {
+        let catalog_json = large_alias_catalog(alias_count);
+        let mut reg = AliasRegistry::new();
+        reg.load_from_json(&catalog_json).unwrap();
+        let resource = large_storage_resource(alias_count);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(alias_count),
+            &(reg, resource),
+            |b, (reg, res)| b.iter(|| normalizer::normalize(black_box(res), Some(reg), None)),
+        );
+    }
+    group.finish();
+}
+
+fn bench_denormalize_large_catalog(c: &mut Criterion) {
+    let mut group = c.benchmark_group("denormalize/large_catalog");
+    for alias_count in [50, 200] {
+        let catalog_json = large_alias_catalog(alias_count);
+        let mut reg = AliasRegistry::new();
+        reg.load_from_json(&catalog_json).unwrap();
+        let resource = large_storage_resource(alias_count);
+        let normalized = normalizer::normalize(&resource, Some(&reg), None);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(alias_count),
+            &(reg, normalized),
+            |b, (reg, norm)| b.iter(|| denormalizer::denormalize(black_box(norm), Some(reg), None)),
+        );
+    }
+    group.finish();
+}
+
+fn bench_nsg_large_subarrays(c: &mut Criterion) {
+    let registry = build_registry();
+    let mut group = c.benchmark_group("round_trip/nsg_sub_resource");
+    for rule_count in [50, 200, 500] {
+        let resource = nsg_resource(rule_count);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(rule_count),
+            &resource,
+            |b, res| {
+                b.iter(|| {
+                    let n = normalizer::normalize(black_box(res), Some(&registry), None);
+                    denormalizer::denormalize(&n, Some(&registry), None)
+                })
+            },
+        );
+    }
+    group.finish();
+}
+
+// ─── Versioned-path benchmarks ──────────────────────────────────────────────
+//
+// Exercise the precomputed versioned-path aggregates by building a catalog
+// where wildcard (array) aliases have version-specific paths that differ from
+// the default, then running normalize/denormalize with an explicit api_version.
+
+/// NSG-like alias catalog where wildcard aliases have versioned paths that
+/// differ from the default.  This forces the normalize/denormalize path through
+/// the versioned aggregate lookup rather than the default-aggregate fast path.
+const VERSIONED_ALIASES_JSON: &str = r#"[
+  {
+    "namespace": "Microsoft.Network",
+    "resourceTypes": [
+      {
+        "resourceType": "networkSecurityGroups",
+        "aliases": [
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].protocol",
+            "defaultPath": "properties.securityRules[*].properties.protocol",
+            "paths": [
+              { "path": "properties.securityRules[*].properties.transportProtocol", "apiVersions": ["2020-01-01"] },
+              { "path": "properties.securityRules[*].properties.protocol", "apiVersions": ["2022-01-01"] }
+            ]
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].access",
+            "defaultPath": "properties.securityRules[*].properties.access",
+            "paths": [
+              { "path": "properties.securityRules[*].properties.accessLevel", "apiVersions": ["2020-01-01"] },
+              { "path": "properties.securityRules[*].properties.access", "apiVersions": ["2022-01-01"] }
+            ]
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].priority",
+            "defaultPath": "properties.securityRules[*].properties.priority",
+            "paths": [
+              { "path": "properties.securityRules[*].properties.rulePriority", "apiVersions": ["2020-01-01"] },
+              { "path": "properties.securityRules[*].properties.priority", "apiVersions": ["2022-01-01"] }
+            ]
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].direction",
+            "defaultPath": "properties.securityRules[*].properties.direction",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].sourceAddressPrefix",
+            "defaultPath": "properties.securityRules[*].properties.sourceAddressPrefix",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].destinationPortRange",
+            "defaultPath": "properties.securityRules[*].properties.destinationPortRange",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/securityRules[*].name",
+            "defaultPath": "properties.securityRules[*].name",
+            "paths": []
+          },
+          {
+            "name": "Microsoft.Network/networkSecurityGroups/provisioningState",
+            "defaultPath": "properties.provisioningState",
+            "paths": [
+              { "path": "properties.state", "apiVersions": ["2020-01-01"] },
+              { "path": "properties.provisioningState", "apiVersions": ["2022-01-01"] }
+            ]
+          }
+        ]
+      }
+    ]
+  }
+]"#;
+
+fn build_versioned_registry() -> AliasRegistry {
+    let mut reg = AliasRegistry::new();
+    reg.load_from_json(VERSIONED_ALIASES_JSON).unwrap();
+    reg
+}
+
+/// Build an NSG resource for versioned-path benchmarks.
+/// Uses the 2020-01-01 field names (`transportProtocol`, `accessLevel`,
+/// `rulePriority`) so that versioned path resolution actually differs from
+/// the default.
+fn nsg_versioned_resource(rule_count: usize) -> Value {
+    let rules: Vec<serde_json::Value> = (0..rule_count)
+        .map(|i| {
+            json!({
+                "name": format!("rule-{}", i),
+                "properties": {
+                    "transportProtocol": "Tcp",
+                    "accessLevel": if i % 2 == 0 { "Allow" } else { "Deny" },
+                    "rulePriority": 100 + i,
+                    "direction": "Inbound",
+                    "sourceAddressPrefix": format!("10.0.{}.0/24", i % 256),
+                    "destinationPortRange": format!("{}", 80 + i)
+                }
+            })
+        })
+        .collect();
+
+    to_regorus(json!({
+        "name": "myNsg",
+        "type": "Microsoft.Network/networkSecurityGroups",
+        "location": "eastus",
+        "properties": {
+            "state": "Succeeded",
+            "securityRules": rules
+        }
+    }))
+}
+
+fn bench_normalize_versioned(c: &mut Criterion) {
+    let registry = build_versioned_registry();
+    let mut group = c.benchmark_group("normalize_versioned/nsg_rules");
+
+    for rule_count in [5, 20, 100] {
+        let resource = nsg_versioned_resource(rule_count);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(rule_count),
+            &resource,
+            |b, res| {
+                b.iter(|| {
+                    normalizer::normalize(black_box(res), Some(&registry), Some("2020-01-01"))
+                })
+            },
+        );
+    }
+    group.finish();
+}
+
+fn bench_denormalize_versioned(c: &mut Criterion) {
+    let registry = build_versioned_registry();
+    let mut group = c.benchmark_group("denormalize_versioned/nsg_rules");
+
+    for rule_count in [5, 20, 100] {
+        let resource = nsg_versioned_resource(rule_count);
+        let normalized = normalizer::normalize(&resource, Some(&registry), Some("2020-01-01"));
+        group.bench_with_input(
+            BenchmarkId::from_parameter(rule_count),
+            &normalized,
+            |b, norm| {
+                b.iter(|| {
+                    denormalizer::denormalize(black_box(norm), Some(&registry), Some("2020-01-01"))
+                })
+            },
+        );
+    }
+    group.finish();
+}
+
+fn bench_round_trip_versioned(c: &mut Criterion) {
+    let registry = build_versioned_registry();
+    let mut group = c.benchmark_group("round_trip_versioned/nsg_rules");
+
+    for rule_count in [20, 100] {
+        let resource = nsg_versioned_resource(rule_count);
+        group.bench_with_input(
+            BenchmarkId::from_parameter(rule_count),
+            &resource,
+            |b, res| {
+                b.iter(|| {
+                    let n =
+                        normalizer::normalize(black_box(res), Some(&registry), Some("2020-01-01"));
+                    denormalizer::denormalize(&n, Some(&registry), Some("2020-01-01"))
+                })
+            },
+        );
+    }
+    group.finish();
+}
+
+criterion_group!(
+    normalization_benches,
+    bench_normalize_simple,
+    bench_normalize_no_aliases,
+    bench_normalize_nsg_scaling,
+    bench_denormalize_simple,
+    bench_denormalize_nsg_scaling,
+    bench_round_trip,
+    bench_normalize_and_wrap,
+    bench_registry_load,
+    bench_normalize_large_catalog,
+    bench_denormalize_large_catalog,
+    bench_nsg_large_subarrays,
+    bench_normalize_versioned,
+    bench_denormalize_versioned,
+    bench_round_trip_versioned,
+);
+criterion_main!(normalization_benches);

benches/regorus_benchmark.rs

@@ -0,0 +1,186 @@
+use std::hint::black_box;
+
+use regorus::{Engine, Value};
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion};
+use serde_json::json;
+
+fn engine_with_policy(policy: &str) -> Engine {
+    let mut engine = Engine::new();
+    engine
+        .add_policy("policy.rego".to_string(), policy.to_string())
+        .unwrap();
+    engine
+}
+
+fn eval_principal(engine: &mut Engine) {
+    engine.set_input(black_box(json!({"principal": "admin"}).into()));
+    let result = engine
+        .eval_rule(black_box("data.bench.allow".to_string()))
+        .unwrap();
+    assert_eq!(result, true.into());
+}
+
+fn allow_with_simple_equality(c: &mut Criterion) {
+    c.bench_function("simple equality check with constant", |b| {
+        let mut engine = engine_with_policy(
+            r#"
+            package bench
+            allow if input.principal == "admin"
+            "#,
+        );
+
+        b.iter(|| eval_principal(&mut engine))
+    });
+
+    c.bench_function("simple equality check with data", |b| {
+        let mut engine = engine_with_policy(
+            r#"
+            package bench
+            allow if input.principal == data.allowed_principal
+            "#,
+        );
+        engine
+            .add_data(json!({"allowed_principal": "admin"}).into())
+            .unwrap();
+
+        b.iter(|| eval_principal(&mut engine))
+    });
+}
+
+fn allow_with_simple_membership(c: &mut Criterion) {
+    let generate_principals = |n: usize| {
+        (0..n)
+            .map(|i| i.to_string())
+            .chain(std::iter::once("admin".to_string()))
+            .collect::<Vec<_>>()
+    };
+
+    let mut group = c.benchmark_group("allow with simple membership");
+    for size in [32, 64, 128, 512, 1024, 2048].iter() {
+        group.bench_with_input(BenchmarkId::new("with constant", size), size, |b, &size| {
+            let principals = generate_principals(size).join("\",\"");
+            let mut engine = engine_with_policy(&format!(
+                r#"
+                package bench
+
+                allowed_principals := {{
+                    "{principals}"
+                }}
+
+                allow if input.principal in allowed_principals
+                "#
+            ));
+
+            b.iter(|| eval_principal(&mut engine))
+        });
+
+        group.bench_with_input(BenchmarkId::new("with data", size), size, |b, &size| {
+            let principals = generate_principals(size);
+            let mut engine = engine_with_policy(
+                r#"
+                package bench
+                allow if input.principal in data.allowed_principals
+                "#,
+            );
+            engine
+                .add_data(json!({"allowed_principals": principals}).into())
+                .unwrap();
+
+            b.iter(|| eval_principal(&mut engine))
+        });
+    }
+    group.finish();
+}
+
+fn clone(c: &mut Criterion) {
+    // Use Arc<BtreeMap> as a reference. Clone will only increment
+    // the reference count.
+    let mut m = std::collections::BTreeMap::default();
+    m.insert(1, 2);
+    let m = std::sync::Arc::new(m);
+
+    c.bench_function("clone: Arc<BTreeMap>", |b| {
+        b.iter(|| {
+            let _ = m.clone();
+        })
+    });
+
+    let mut engine = Engine::new();
+    engine.set_rego_v0(true);
+    engine
+        .add_policy_from_file("tests/aci/framework.rego")
+        .unwrap();
+    engine.add_policy_from_file("tests/aci/api.rego").unwrap();
+    engine
+        .add_policy_from_file("tests/aci/policy.rego")
+        .unwrap();
+    engine
+        .add_data(Value::from_json_file("tests/aci/data.json").expect("failed to load data.json"))
+        .expect("failed to add data");
+    engine.set_input(
+        Value::from_json_file("tests/aci/input.json").expect("failed to load input.json"),
+    );
+
+    // An engine without preparation will not have processed fields populated.
+    c.bench_function("clone: engine with aci policies", |b| {
+        b.iter(|| {
+            let _ = engine.clone();
+        })
+    });
+
+    // Trigger engine preparation.
+    let _ = engine.eval_query("data.framework.mount_overlay".to_string(), false);
+
+    // Prepared engine will have many more fields populated. But the fields are
+    // immutable after preparation and will be shared between clones.
+    c.bench_function("clone: prepared engine with aci policies", |b| {
+        b.iter(|| {
+            let _ = engine.clone();
+        })
+    });
+}
+
+fn aci_policy_eval(c: &mut Criterion) {
+    let mut group = c.benchmark_group("ACI Policy Eval");
+    let rules = ["data.policy.mount_overlay", "data.policy.mount_device"];
+    for rule in rules {
+        group.bench_with_input(BenchmarkId::new("rule", rule), &rule, |b, rule| {
+            let mut engine = Engine::new();
+            engine.set_rego_v0(true);
+
+            engine
+                .add_policy_from_file("tests/aci/api.rego")
+                .expect("failed to add api.rego");
+            engine
+                .add_policy_from_file("tests/aci/framework.rego")
+                .expect("failed to add framework.rego");
+            engine
+                .add_policy_from_file("tests/aci/policy.rego")
+                .expect("failed to add policy.rego");
+            engine
+                .add_data(
+                    Value::from_json_file("tests/aci/data.json").expect("failed to load data.json"),
+                )
+                .expect("failed to add data");
+            let input =
+                Value::from_json_file("tests/aci/input.json").expect("failed to load input.json");
+            engine.set_input(input.clone());
+            engine.eval_rule(rule.to_string()).unwrap();
+            b.iter(|| {
+                engine.eval_rule(rule.to_string()).unwrap();
+            })
+        });
+    }
+    group.finish();
+}
+
+criterion_group!(
+    benches,
+    allow_with_simple_equality,
+    allow_with_simple_membership,
+    clone,
+    aci_policy_eval
+);
+
+criterion_main!(benches);

benches/rvm_benchmark.rs

@@ -0,0 +1,680 @@
+// Copyright (c) Microsoft Corporation.
+// Licensed under the MIT License.
+
+//! Comprehensive RVM benchmarks covering all aspects of the Rego Virtual Machine.
+//!
+//! # Policy families
+//!
+//! | Family     | Source                        | Policies | Inputs/policy |
+//! |------------|-------------------------------|----------|---------------|
+//! | Synthetic  | `benches/evaluation/test_data`| 9        | 3 each        |
+//! | ACI        | `tests/aci`                   | 9        | 1 each        |
+//!
+//! # Benchmark groups
+//!
+//! | Group                    | What it measures                                      |
+//! |--------------------------|-------------------------------------------------------|
+//! | `cold/{case}/{config}`   | Cold: new VM + load + data + input + execute          |
+//! | `hot/{case}/{config}`    | Hot: set_input + execute (VM reused across iters)     |
+//! | `compilation`            | Rego CompiledPolicy → RVM Program                     |
+//! | `serialization`          | Program binary serialize / deserialize roundtrip       |
+//! | `startup`                | Isolated VM creation & setup overhead                 |
+//! | `stats`                  | Instruction/literal counts (reported as throughput)    |
+//! | `end_to_end`             | Full roundtrip: compile → serialize → deserialize → eval |
+//!
+//! # Running subsets
+//!
+//! ```sh
+//! cargo bench --bench rvm_benchmark                              # everything
+//! cargo bench --bench rvm_benchmark -- cold                      # all cold eval
+//! cargo bench --bench rvm_benchmark -- hot                       # all hot eval
+//! cargo bench --bench rvm_benchmark -- regular_with_limits       # one config across cases
+//! cargo bench --bench rvm_benchmark -- cold/aci/                 # all ACI cold benchmarks
+//! cargo bench --bench rvm_benchmark -- rbac                      # one policy family
+//! cargo bench --bench rvm_benchmark -- compilation               # compilation only
+//! cargo bench --bench rvm_benchmark -- serialization             # serialization only
+//! cargo bench --bench rvm_benchmark -- startup                   # startup overhead
+//! ```
+
+use std::hint::black_box;
+use std::num::NonZeroU32;
+use std::path::Path;
+use std::sync::Arc;
+use std::time::Duration;
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion, Throughput};
+use serde::{Deserialize, Serialize};
+use walkdir::WalkDir;
+
+use regorus::languages::rego::compiler::Compiler;
+use regorus::rvm::program::Program;
+use regorus::rvm::vm::{ExecutionMode, RegoVM};
+use regorus::utils::limits::ExecutionTimerConfig;
+use regorus::{Engine, Rc, Value};
+
+// ---------------------------------------------------------------------------
+// Limit constants – generous ceilings that still exercise the limit-checking
+// hot path (memory_check, execution_timer_tick, instruction-limit compare).
+// ---------------------------------------------------------------------------
+
+#[cfg(feature = "allocator-memory-limits")]
+const MEMORY_LIMIT_BYTES: u64 = 256 * 1024 * 1024;
+const TIME_LIMIT: Duration = Duration::from_secs(30);
+const TIMER_CHECK_INTERVAL: NonZeroU32 = NonZeroU32::new(16).unwrap();
+const INSTRUCTION_LIMIT: usize = 10_000_000;
+
+#[derive(Clone, Copy)]
+struct EvalConfig {
+    name: &'static str,
+    mode: ExecutionMode,
+    limits: bool,
+}
+
+const EVAL_CONFIGS: [EvalConfig; 4] = [
+    EvalConfig {
+        name: "regular_no_limits",
+        mode: ExecutionMode::RunToCompletion,
+        limits: false,
+    },
+    EvalConfig {
+        name: "regular_with_limits",
+        mode: ExecutionMode::RunToCompletion,
+        limits: true,
+    },
+    EvalConfig {
+        name: "suspendable_no_limits",
+        mode: ExecutionMode::Suspendable,
+        limits: false,
+    },
+    EvalConfig {
+        name: "suspendable_with_limits",
+        mode: ExecutionMode::Suspendable,
+        limits: true,
+    },
+];
+
+// ---------------------------------------------------------------------------
+// Data types
+// ---------------------------------------------------------------------------
+
+/// A compiled benchmark program ready for RVM execution.
+struct BenchmarkProgram {
+    /// Human-readable name (e.g. "rbac_policy" or "aci/create_container").
+    name: String,
+    /// Pre-compiled RVM program.
+    program: Arc<Program>,
+    /// Compiled policy (kept for compilation benchmarks).
+    compiled_policy: regorus::CompiledPolicy,
+    /// Entry-point rule path.
+    entry_point: String,
+    /// Data object (Some for policies that require external data like ACI).
+    data: Option<Value>,
+    /// Named inputs for this policy.
+    inputs: Vec<(String, Value)>,
+}
+
+// ---------------------------------------------------------------------------
+// ACI YAML types
+// ---------------------------------------------------------------------------
+
+#[derive(Serialize, Deserialize, Debug)]
+struct AciTestCase {
+    note: String,
+    data: Value,
+    input: Value,
+    modules: Vec<String>,
+    query: String,
+    want_result: Value,
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+struct AciYamlTest {
+    cases: Vec<AciTestCase>,
+}
+
+// ---------------------------------------------------------------------------
+// Synthetic policy loading
+// ---------------------------------------------------------------------------
+
+/// Policy ↔ input file mapping for synthetic policies.
+const SYNTHETIC_POLICIES: &[(&str, &str, &[&str])] = &[
+    (
+        "rbac_policy",
+        "rbac_policy.rego",
+        &["rbac_input.json", "rbac_input2.json", "rbac_input3.json"],
+    ),
+    (
+        "api_access",
+        "api_access_policy.rego",
+        &[
+            "api_access_input.json",
+            "api_access_input2.json",
+            "api_access_input3.json",
+        ],
+    ),
+    (
+        "data_sensitivity",
+        "data_sensitivity_policy.rego",
+        &[
+            "data_sensitivity_input.json",
+            "data_sensitivity_input2.json",
+            "data_sensitivity_input3.json",
+        ],
+    ),
+    (
+        "time_based",
+        "time_based_policy.rego",
+        &[
+            "time_based_input.json",
+            "time_based_input2.json",
+            "time_based_input3.json",
+        ],
+    ),
+    (
+        "data_processing",
+        "data_processing_policy.rego",
+        &[
+            "data_processing_input.json",
+            "data_processing_input2.json",
+            "data_processing_input3.json",
+        ],
+    ),
+    (
+        "azure_vm",
+        "azure_vm_policy.rego",
+        &[
+            "azure_vm_input.json",
+            "azure_vm_input2.json",
+            "azure_vm_input3.json",
+        ],
+    ),
+    (
+        "azure_storage",
+        "azure_storage_policy.rego",
+        &[
+            "azure_storage_input.json",
+            "azure_storage_input2.json",
+            "azure_storage_input3.json",
+        ],
+    ),
+    (
+        "azure_keyvault",
+        "azure_keyvault_policy.rego",
+        &[
+            "azure_keyvault_input.json",
+            "azure_keyvault_input2.json",
+            "azure_keyvault_input3.json",
+        ],
+    ),
+    (
+        "azure_nsg",
+        "azure_nsg_policy.rego",
+        &[
+            "azure_nsg_input.json",
+            "azure_nsg_input2.json",
+            "azure_nsg_input3.json",
+        ],
+    ),
+];
+
+/// Compile synthetic Rego policies into RVM programs.
+fn compile_synthetic_programs() -> Vec<BenchmarkProgram> {
+    let base_dir = Path::new(env!("CARGO_MANIFEST_DIR"))
+        .join("benches")
+        .join("evaluation")
+        .join("test_data");
+
+    let entry_point = "data.bench.allow";
+    let entry_point_rc: Rc<str> = entry_point.into();
+
+    SYNTHETIC_POLICIES
+        .iter()
+        .map(|(name, policy_file, input_files)| {
+            let policy_path = base_dir.join("policies").join(policy_file);
+            let policy_content = std::fs::read_to_string(&policy_path)
+                .unwrap_or_else(|e| panic!("Failed to read {policy_path:?}: {e}"));
+
+            let mut engine = Engine::new();
+            engine
+                .add_policy("policy.rego".to_string(), policy_content)
+                .expect("failed to add policy");
+
+            let compiled_policy = engine
+                .compile_with_entrypoint(&entry_point_rc)
+                .expect("failed to compile policy");
+
+            let program = Compiler::compile_from_policy(&compiled_policy, &[entry_point])
+                .expect("failed to compile to RVM program");
+
+            let inputs: Vec<(String, Value)> = input_files
+                .iter()
+                .map(|input_file| {
+                    let input_path = base_dir.join("inputs").join(input_file);
+                    let json = std::fs::read_to_string(&input_path)
+                        .unwrap_or_else(|e| panic!("Failed to read {input_path:?}: {e}"));
+                    let value = Value::from_json_str(&json).expect("failed to parse input JSON");
+                    let display = input_file.trim_end_matches(".json").to_string();
+                    (display, value)
+                })
+                .collect();
+
+            BenchmarkProgram {
+                name: name.to_string(),
+                program,
+                compiled_policy,
+                entry_point: entry_point.to_string(),
+                data: None,
+                inputs,
+            }
+        })
+        .collect()
+}
+
+// ---------------------------------------------------------------------------
+// ACI policy loading
+// ---------------------------------------------------------------------------
+
+/// Load all ACI test cases from YAML files.
+fn load_aci_cases(dir: &Path) -> Vec<AciTestCase> {
+    let mut cases = Vec::new();
+    for entry in WalkDir::new(dir)
+        .sort_by_file_name()
+        .into_iter()
+        .filter_map(|e| e.ok())
+    {
+        let path = entry.path();
+        if !path.to_string_lossy().ends_with(".yaml") {
+            continue;
+        }
+        let yaml = std::fs::read(path).expect("failed to read yaml");
+        let yaml = String::from_utf8_lossy(&yaml);
+        let test: AciYamlTest = serde_yaml::from_str(&yaml).expect("failed to deserialize yaml");
+        cases.extend(test.cases);
+    }
+    cases
+}
+
+/// Build an Engine with policies loaded for a given ACI test case.
+fn build_aci_engine(dir: &Path, case: &AciTestCase) -> Engine {
+    let mut engine = Engine::new();
+    engine.set_rego_v0(true);
+    engine
+        .add_data(case.data.clone())
+        .expect("failed to add data");
+    engine.set_input(case.input.clone());
+    for (idx, rego) in case.modules.iter().enumerate() {
+        if rego.ends_with(".rego") {
+            engine
+                .add_policy_from_file(dir.join(rego).to_str().expect("invalid path"))
+                .expect("failed to add policy");
+        } else {
+            engine
+                .add_policy(format!("rego{idx}.rego"), rego.clone())
+                .expect("failed to add policy");
+        }
+    }
+    engine
+}
+
+/// Compile ACI test cases into RVM programs.
+fn compile_aci_programs() -> Vec<BenchmarkProgram> {
+    let dir = Path::new("tests/aci");
+    load_aci_cases(dir)
+        .into_iter()
+        .map(|case| {
+            let mut engine = build_aci_engine(dir, &case);
+            let rule = case.query.replace("=x", "");
+            let rule_rc: Rc<str> = rule.clone().into();
+            let compiled_policy = engine
+                .compile_with_entrypoint(&rule_rc)
+                .expect("failed to compile");
+            let program = Compiler::compile_from_policy(&compiled_policy, &[rule.as_str()])
+                .expect("failed to compile to RVM");
+
+            BenchmarkProgram {
+                name: format!("aci/{}", case.note),
+                program,
+                compiled_policy,
+                entry_point: rule,
+                data: Some(case.data),
+                inputs: vec![("input".to_string(), case.input)],
+            }
+        })
+        .collect()
+}
+
+// ---------------------------------------------------------------------------
+// Compile all policies
+// ---------------------------------------------------------------------------
+
+/// Compile all policies (synthetic + ACI) into RVM programs.
+fn compile_all_programs() -> Vec<BenchmarkProgram> {
+    let mut programs = compile_synthetic_programs();
+    programs.extend(compile_aci_programs());
+    programs
+}
+
+// ---------------------------------------------------------------------------
+// Limit helpers
+// ---------------------------------------------------------------------------
+
+/// Apply or remove production-style limits based on a boolean flag.
+fn configure_limits(vm: &mut RegoVM, limits: bool) {
+    if limits {
+        #[cfg(feature = "allocator-memory-limits")]
+        regorus::set_global_memory_limit(Some(MEMORY_LIMIT_BYTES));
+        vm.set_execution_timer_config(Some(ExecutionTimerConfig {
+            limit: TIME_LIMIT,
+            check_interval: TIMER_CHECK_INTERVAL,
+        }));
+        vm.set_max_instructions(INSTRUCTION_LIMIT);
+    } else {
+        #[cfg(feature = "allocator-memory-limits")]
+        regorus::set_global_memory_limit(None);
+        vm.set_execution_timer_config(None);
+        vm.set_max_instructions(usize::MAX);
+    }
+}
+
+// ---------------------------------------------------------------------------
+// Cold evaluation — new VM per iteration (full setup + execute)
+//
+// Benchmarks are registered case-first so each workload is shown with all
+// config variants adjacent to one another, making per-case comparisons easier.
+// ---------------------------------------------------------------------------
+
+fn bench_cold(c: &mut Criterion) {
+    let programs = compile_all_programs();
+    let mut group = c.benchmark_group("cold");
+
+    for bp in &programs {
+        for (input_name, input_value) in &bp.inputs {
+            let case_id = if bp.inputs.len() == 1 {
+                bp.name.clone()
+            } else {
+                format!("{}/{}", bp.name, input_name)
+            };
+            let program = bp.program.clone();
+            let data = bp.data.clone();
+            let input = input_value.clone();
+
+            for config in EVAL_CONFIGS {
+                group.bench_function(BenchmarkId::new(&case_id, config.name), |b| {
+                    b.iter(|| {
+                        let mut vm = RegoVM::new();
+                        vm.set_execution_mode(config.mode);
+                        vm.load_program(black_box(program.clone()));
+                        if let Some(ref d) = data {
+                            vm.set_data(black_box(d.clone())).unwrap();
+                        }
+                        vm.set_input(black_box(input.clone()));
+                        configure_limits(&mut vm, config.limits);
+                        black_box(vm.execute().unwrap())
+                    })
+                });
+            }
+        }
+    }
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// Hot evaluation — VM reused across iterations
+//
+// The VM is created once with program, data, mode, and limits.  Each
+// iteration only calls set_input + execute, measuring pure execution
+// overhead with minimal setup.  A warm-up execution fills the register
+// window pool so all iterations benefit from pooled allocations.
+// ---------------------------------------------------------------------------
+
+fn bench_hot(c: &mut Criterion) {
+    let programs = compile_all_programs();
+    let mut group = c.benchmark_group("hot");
+
+    for bp in &programs {
+        let program = bp.program.clone();
+        let data = bp.data.clone();
+        let inputs: Vec<Value> = bp.inputs.iter().map(|(_, v)| v.clone()).collect();
+        let num_inputs = inputs.len();
+
+        for config in EVAL_CONFIGS {
+            group.bench_function(BenchmarkId::new(&bp.name, config.name), |b| {
+                let mut vm = RegoVM::new();
+                vm.set_execution_mode(config.mode);
+                vm.load_program(program.clone());
+                if let Some(ref d) = data {
+                    vm.set_data(d.clone()).unwrap();
+                }
+                configure_limits(&mut vm, config.limits);
+
+                // Warm up: fill register window pools, caches, etc.
+                vm.set_input(inputs[0].clone());
+                vm.execute().expect("warm-up failed");
+
+                let mut i = 0usize;
+                b.iter(|| {
+                    let input = &inputs[i % num_inputs];
+                    vm.set_input(black_box(input.clone()));
+                    black_box(vm.execute().unwrap());
+                    i += 1;
+                })
+            });
+        }
+    }
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// Compilation — Rego CompiledPolicy → RVM Program
+// ---------------------------------------------------------------------------
+
+fn bench_compilation(c: &mut Criterion) {
+    let programs = compile_all_programs();
+    let mut group = c.benchmark_group("compilation");
+
+    for bp in &programs {
+        let entry_point: &str = &bp.entry_point;
+        group.bench_with_input(
+            BenchmarkId::new("rego_to_rvm", &bp.name),
+            &bp.compiled_policy,
+            |b, compiled_policy| {
+                b.iter(|| {
+                    Compiler::compile_from_policy(
+                        black_box(compiled_policy),
+                        black_box(&[entry_point]),
+                    )
+                    .unwrap();
+                })
+            },
+        );
+    }
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// Serialization — binary serialize / deserialize roundtrip
+// ---------------------------------------------------------------------------
+
+fn bench_serialization(c: &mut Criterion) {
+    let programs = compile_all_programs();
+    let mut group = c.benchmark_group("serialization");
+
+    for bp in &programs {
+        let program = &bp.program;
+        let serialized = program
+            .serialize_binary()
+            .expect("failed to serialize program");
+        let byte_len = serialized.len() as u64;
+
+        group.throughput(Throughput::Bytes(byte_len));
+        group.bench_function(BenchmarkId::new("serialize", &bp.name), |b| {
+            b.iter(|| black_box(program.serialize_binary().unwrap()))
+        });
+
+        group.throughput(Throughput::Bytes(byte_len));
+        group.bench_function(BenchmarkId::new("deserialize", &bp.name), |b| {
+            b.iter(|| black_box(Program::deserialize_binary(black_box(&serialized)).unwrap()))
+        });
+    }
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// Startup — isolated VM creation & setup overhead
+// ---------------------------------------------------------------------------
+
+fn bench_startup(c: &mut Criterion) {
+    let programs = compile_all_programs();
+    let mut group = c.benchmark_group("startup");
+
+    // Use the first program as representative for startup overhead.
+    let bp = &programs[0];
+    let program = bp.program.clone();
+    let input = bp.inputs[0].1.clone();
+
+    // Bare VM creation
+    group.bench_function("new", |b| b.iter(|| black_box(RegoVM::new())));
+
+    // load_program (Arc clone + internal setup)
+    group.bench_function("load_program", |b| {
+        b.iter(|| {
+            let mut vm = RegoVM::new();
+            vm.load_program(black_box(program.clone()));
+            black_box(&vm);
+        })
+    });
+
+    // set_input
+    group.bench_function("set_input", |b| {
+        let mut vm = RegoVM::new();
+        vm.load_program(program.clone());
+        b.iter(|| {
+            vm.set_input(black_box(input.clone()));
+        })
+    });
+
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// Stats — instruction / literal counts (reported as throughput)
+// ---------------------------------------------------------------------------
+
+fn bench_stats(c: &mut Criterion) {
+    let programs = compile_all_programs();
+
+    eprintln!();
+    eprintln!(
+        "{:<30} {:>8} {:>8} {:>8} {:>10}",
+        "program", "instrs", "lits", "entries", "bytes"
+    );
+    eprintln!("{}", "-".repeat(70));
+
+    let mut group = c.benchmark_group("stats");
+    for bp in &programs {
+        let serialized = bp.program.serialize_binary().expect("serialize failed");
+        let byte_len = serialized.len();
+        let instr_count = bp.program.instructions.len();
+        let lit_count = bp.program.literals.len();
+        let entry_count = bp.program.entry_points.len();
+
+        eprintln!(
+            "{:<30} {:>8} {:>8} {:>8} {:>10}",
+            bp.name, instr_count, lit_count, entry_count, byte_len,
+        );
+
+        group.throughput(Throughput::Elements(instr_count as u64));
+        group.bench_function(BenchmarkId::new("serialize", &bp.name), |b| {
+            b.iter(|| black_box(bp.program.serialize_binary().unwrap()))
+        });
+    }
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// End-to-end roundtrip (compile + serialize + deserialize + eval)
+//
+// Only runs for synthetic policies where we have direct access to rego
+// source files.  ACI policies are loaded from YAML with module references
+// which makes the setup pipeline different.
+// ---------------------------------------------------------------------------
+
+fn bench_end_to_end(c: &mut Criterion) {
+    let base_dir = Path::new(env!("CARGO_MANIFEST_DIR"))
+        .join("benches")
+        .join("evaluation")
+        .join("test_data");
+
+    let entry_point = "data.bench.allow";
+    let entry_point_rc: Rc<str> = entry_point.into();
+
+    let mut group = c.benchmark_group("end_to_end");
+
+    for &(name, policy_file, input_files) in SYNTHETIC_POLICIES {
+        let policy_path = base_dir.join("policies").join(policy_file);
+        let policy_content = std::fs::read_to_string(&policy_path)
+            .unwrap_or_else(|e| panic!("Failed to read {policy_path:?}: {e}"));
+
+        // Use just the first input for end-to-end
+        let input_path = base_dir.join("inputs").join(input_files[0]);
+        let input_json = std::fs::read_to_string(&input_path)
+            .unwrap_or_else(|e| panic!("Failed to read {input_path:?}: {e}"));
+
+        group.bench_function(BenchmarkId::new("roundtrip", name), |b| {
+            b.iter(|| {
+                // 1. Engine + parse
+                let mut engine = Engine::new();
+                engine
+                    .add_policy("policy.rego".to_string(), policy_content.clone())
+                    .unwrap();
+
+                // 2. Compile to CompiledPolicy
+                let compiled_policy = engine.compile_with_entrypoint(&entry_point_rc).unwrap();
+
+                // 3. Compile to RVM Program
+                let program =
+                    Compiler::compile_from_policy(&compiled_policy, &[entry_point]).unwrap();
+
+                // 4. Serialize
+                let bytes = program.serialize_binary().unwrap();
+
+                // 5. Deserialize
+                let deserialized = Program::deserialize_binary(&bytes).unwrap();
+                let program = match deserialized {
+                    regorus::rvm::program::DeserializationResult::Complete(p) => Arc::new(p),
+                    regorus::rvm::program::DeserializationResult::Partial(p) => {
+                        Arc::new(Program::compile_from_partial(p).unwrap())
+                    }
+                };
+
+                // 6. Execute
+                let mut vm = RegoVM::new();
+                vm.load_program(program);
+                let input = Value::from_json_str(&input_json).unwrap();
+                vm.set_input(input);
+                black_box(vm.execute().unwrap());
+            })
+        });
+    }
+
+    group.finish();
+}
+
+// ---------------------------------------------------------------------------
+// Criterion groups — organised for selective runs
+// ---------------------------------------------------------------------------
+
+criterion_group!(cold_benches, bench_cold);
+
+criterion_group!(hot_benches, bench_hot);
+
+criterion_group!(
+    misc_benches,
+    bench_compilation,
+    bench_serialization,
+    bench_startup,
+    bench_stats,
+    bench_end_to_end,
+);
+
+criterion_main!(cold_benches, hot_benches, misc_benches);

benches/schema_validation_benchmark.rs

@@ -0,0 +1,886 @@
+use criterion::{criterion_group, criterion_main, Criterion};
+use regorus::Value;
+use regorus::{Schema, SchemaValidator};
+use serde_json::json;
+
+// Observed: validate_string - 3.19 ns/iter
+fn bench_string_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "string",
+        "minLength": 3,
+        "maxLength": 10
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from("hello");
+
+    c.bench_function("validate_string", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_number - 146.5 ns/iter
+fn bench_number_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "number",
+        "minimum": 0.0,
+        "maximum": 100.0
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(42.5);
+
+    c.bench_function("validate_number", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_array - 95.0 ns/iter
+fn bench_array_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "array",
+        "items": { "type": "integer" },
+        "minItems": 2,
+        "maxItems": 5
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!([1, 2, 3]));
+
+    c.bench_function("validate_array", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_object - 126.9 ns/iter
+fn bench_object_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "name": { "type": "string" },
+            "age": { "type": "integer", "minimum": 0 }
+        },
+        "required": ["name", "age"]
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({"name": "Alice", "age": 30}));
+
+    c.bench_function("validate_object", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_complex_nested - 710.5 ns/iter
+fn bench_complex_nested_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "user": {
+                "type": "object",
+                "properties": {
+                    "id": { "type": "string" },
+                    "profile": {
+                        "type": "object",
+                        "properties": {
+                            "email": { "type": "string" },
+                            "roles": {
+                                "type": "array",
+                                "items": { "type": "string" }
+                            }
+                        },
+                        "required": ["email", "roles"]
+                    }
+                },
+                "required": ["id", "profile"]
+            },
+            "active": { "type": "boolean" }
+        },
+        "required": ["user", "active"]
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "user": {
+            "id": "u123",
+            "profile": {
+                "email": "alice@example.com",
+                "roles": ["admin", "user"]
+            }
+        },
+        "active": true
+    }));
+
+    c.bench_function("validate_complex_nested", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_string_pattern - 29.99 µs/iter
+fn bench_string_pattern_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "string",
+        "pattern": "^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}$"
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from("user@example.com");
+
+    c.bench_function("validate_string_pattern", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_enum - 7.26 ns/iter
+fn bench_enum_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "enum": ["pending", "approved", "rejected", "cancelled"]
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from("approved");
+
+    c.bench_function("validate_enum", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_boolean - 3.22 ns/iter
+fn bench_boolean_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "boolean"
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(true);
+
+    c.bench_function("validate_boolean", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_null - 3.22 ns/iter
+fn bench_null_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "null"
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::Null;
+
+    c.bench_function("validate_null", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_large_array - 17.30 µs/iter
+fn bench_large_array_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "array",
+        "items": { "type": "number" },
+        "minItems": 50,
+        "maxItems": 200
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let large_array: Vec<_> = (0..100).map(|i| json!(i as f64)).collect();
+    let value = Value::from(json!(large_array));
+
+    c.bench_function("validate_large_array", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_deeply_nested - 468.2 ns/iter
+fn bench_deeply_nested_object(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "level1": {
+                "type": "object",
+                "properties": {
+                    "level2": {
+                        "type": "object",
+                        "properties": {
+                            "level3": {
+                                "type": "object",
+                                "properties": {
+                                    "level4": {
+                                        "type": "object",
+                                        "properties": {
+                                            "level5": {
+                                                "type": "string"
+                                            }
+                                        },
+                                        "required": ["level5"]
+                                    }
+                                },
+                                "required": ["level4"]
+                            }
+                        },
+                        "required": ["level3"]
+                    }
+                },
+                "required": ["level2"]
+            }
+        },
+        "required": ["level1"]
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "level1": {
+            "level2": {
+                "level3": {
+                    "level4": {
+                        "level5": "deep value"
+                    }
+                }
+            }
+        }
+    }));
+
+    c.bench_function("validate_deeply_nested", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_mixed_type_array - 1.36 µs/iter
+fn bench_mixed_type_array(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "array",
+        "items": {
+            "anyOf": [
+                { "type": "string" },
+                { "type": "number" },
+                { "type": "boolean" }
+            ]
+        }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!(["hello", 42, true, "world", 99.5, false]));
+
+    c.bench_function("validate_mixed_type_array", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_additional_properties - 366.4 ns/iter
+fn bench_additional_properties(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "name": { "type": "string" },
+            "age": { "type": "integer" }
+        },
+        "additionalProperties": { "type": "string" },
+        "required": ["name"]
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "name": "Alice",
+        "age": 30,
+        "city": "New York",
+        "country": "USA",
+        "occupation": "Engineer"
+    }));
+
+    c.bench_function("validate_additional_properties", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_array_constraints - 146.2 ns/iter
+fn bench_array_constraints(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "array",
+        "items": { "type": "string" },
+        "minItems": 2,
+        "maxItems": 10
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!(["apple", "banana", "cherry", "date", "elderberry"]));
+
+    c.bench_function("validate_array_constraints", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_multi_level - 915.8 ns/iter
+fn bench_multi_level_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "user": {
+                "type": "object",
+                "properties": {
+                    "profile": {
+                        "type": "object",
+                        "properties": {
+                            "settings": {
+                                "type": "object",
+                                "properties": {
+                                    "theme": {
+                                        "enum": ["light", "dark", "auto"]
+                                    },
+                                    "notifications": {
+                                        "type": "boolean"
+                                    }
+                                },
+                                "required": ["theme"],
+                                "additionalProperties": { "type": "string" }
+                            }
+                        },
+                        "required": ["settings"],
+                        "additionalProperties": { "type": "any" }
+                    }
+                },
+                "required": ["profile"],
+                "additionalProperties": { "type": "any" }
+            }
+        },
+        "required": ["user"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "user": {
+            "profile": {
+                "settings": {
+                    "theme": "dark",
+                    "notifications": true,
+                    "language": "en"
+                },
+                "avatar": "default.png"
+            },
+            "lastLogin": "2024-01-01"
+        },
+        "metadata": "extra info"
+    }));
+
+    c.bench_function("validate_multi_level", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Azure Resource Validation Benchmarks
+
+// Observed: validate_azure_vm_resource - 34.74 µs/iter
+fn bench_azure_vm_resource_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "type": {
+                "const": "Microsoft.Compute/virtualMachines"
+            },
+            "apiVersion": {
+                "enum": ["2021-03-01", "2021-07-01", "2022-03-01"]
+            },
+            "name": {
+                "type": "string",
+                "pattern": "^[a-zA-Z0-9-._]{1,64}$"
+            },
+            "location": {
+                "type": "string",
+                "description": "Azure region where the VM will be deployed"
+            },
+            "properties": {
+                "type": "object",
+                "properties": {
+                    "hardwareProfile": {
+                        "type": "object",
+                        "properties": {
+                            "vmSize": {
+                                "enum": ["Standard_B1s", "Standard_B2s", "Standard_D2s_v3", "Standard_D4s_v3"]
+                            }
+                        },
+                        "required": ["vmSize"]
+                    },
+                    "osProfile": {
+                        "type": "object",
+                        "properties": {
+                            "computerName": {
+                                "type": "string"
+                            },
+                            "adminUsername": {
+                                "type": "string"
+                            }
+                        },
+                        "required": ["computerName", "adminUsername"]
+                    }
+                },
+                "required": ["hardwareProfile", "osProfile"]
+            }
+        },
+        "required": ["type", "apiVersion", "name", "location", "properties"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "type": "Microsoft.Compute/virtualMachines",
+        "apiVersion": "2021-03-01",
+        "name": "my-vm-01",
+        "location": "eastus",
+        "properties": {
+            "hardwareProfile": {
+                "vmSize": "Standard_B2s"
+            },
+            "osProfile": {
+                "computerName": "my-computer",
+                "adminUsername": "azureuser"
+            }
+        }
+    }));
+
+    c.bench_function("validate_azure_vm_resource", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_azure_storage_resource - 22.12 µs/iter
+fn bench_azure_storage_resource_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "type": {
+                "const": "Microsoft.Storage/storageAccounts"
+            },
+            "apiVersion": {
+                "enum": ["2021-04-01", "2021-06-01", "2022-05-01"]
+            },
+            "name": {
+                "type": "string",
+                "pattern": "^[a-z0-9]{3,24}$"
+            },
+            "location": {
+                "type": "string"
+            },
+            "sku": {
+                "type": "object",
+                "properties": {
+                    "name": {
+                        "enum": ["Standard_LRS", "Standard_GRS", "Standard_RAGRS", "Premium_LRS"]
+                    }
+                },
+                "required": ["name"]
+            },
+            "kind": {
+                "enum": ["Storage", "StorageV2", "BlobStorage", "FileStorage", "BlockBlobStorage"]
+            },
+            "properties": {
+                "type": "object",
+                "properties": {
+                    "accessTier": {
+                        "enum": ["Hot", "Cool", "Archive"]
+                    },
+                    "encryption": {
+                        "type": "object",
+                        "properties": {
+                            "services": {
+                                "type": "object"
+                            }
+                        }
+                    }
+                },
+                "additionalProperties": { "type": "any" }
+            }
+        },
+        "required": ["type", "apiVersion", "name", "location", "sku", "kind"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "type": "Microsoft.Storage/storageAccounts",
+        "apiVersion": "2021-04-01",
+        "name": "mystorageaccount001",
+        "location": "westus2",
+        "sku": {
+            "name": "Standard_LRS"
+        },
+        "kind": "StorageV2",
+        "properties": {
+            "accessTier": "Hot",
+            "encryption": {
+                "services": {}
+            }
+        }
+    }));
+
+    c.bench_function("validate_azure_storage_resource", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_azure_arm_template - 1.99 µs/iter
+fn bench_azure_arm_template_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "$schema": {
+                "type": "string"
+            },
+            "contentVersion": {
+                "type": "string"
+            },
+            "parameters": {
+                "type": "object",
+                "additionalProperties": { "type": "any" }
+            },
+            "variables": {
+                "type": "object",
+                "additionalProperties": { "type": "any" }
+            },
+            "resources": {
+                "type": "array",
+                "items": {
+                    "type": "object",
+                    "properties": {
+                        "type": {
+                            "type": "string"
+                        },
+                        "apiVersion": {
+                            "type": "string"
+                        },
+                        "name": {
+                            "type": "string"
+                        },
+                        "location": {
+                            "type": "string"
+                        },
+                        "properties": {
+                            "type": "object",
+                            "additionalProperties": { "type": "any" }
+                        },
+                        "tags": {
+                            "type": "object",
+                            "additionalProperties": { "type": "string" }
+                        }
+                    },
+                    "required": ["type", "apiVersion", "name"],
+                    "additionalProperties": { "type": "any" }
+                }
+            },
+            "outputs": {
+                "type": "object",
+                "additionalProperties": { "type": "any" }
+            }
+        },
+        "required": ["resources"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "$schema": "https://schema.management.azure.com/schemas/2019-04-01/deploymentTemplate.json#",
+        "contentVersion": "1.0.0.0",
+        "parameters": {
+            "vmName": {
+                "type": "string",
+                "defaultValue": "myVM"
+            }
+        },
+        "variables": {
+            "storageAccountName": "[concat('storage', uniqueString(resourceGroup().id))]"
+        },
+        "resources": [
+            {
+                "type": "Microsoft.Compute/virtualMachines",
+                "apiVersion": "2021-03-01",
+                "name": "[parameters('vmName')]",
+                "location": "[resourceGroup().location]",
+                "properties": {
+                    "hardwareProfile": {
+                        "vmSize": "Standard_B1s"
+                    }
+                },
+                "tags": {
+                    "environment": "dev",
+                    "project": "test"
+                }
+            }
+        ],
+        "outputs": {
+            "vmId": {
+                "type": "string",
+                "value": "[resourceId('Microsoft.Compute/virtualMachines', parameters('vmName'))]"
+            }
+        }
+    }));
+
+    c.bench_function("validate_azure_arm_template", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Azure Policy Effect Validation Benchmarks
+
+// Observed: validate_azure_policy_deny_effect - 188.6 ns/iter
+fn bench_azure_policy_deny_effect_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "effect": {
+                "const": "deny"
+            },
+            "description": {
+                "type": "string"
+            }
+        },
+        "required": ["effect"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "effect": "deny",
+        "description": "Deny resources that don't meet security requirements"
+    }));
+
+    c.bench_function("validate_azure_policy_deny_effect", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_azure_policy_audit_effect - 516.7 ns/iter
+fn bench_azure_policy_audit_effect_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "effect": {
+                "const": "audit"
+            },
+            "description": {
+                "type": "string"
+            },
+            "auditDetails": {
+                "type": "object",
+                "properties": {
+                    "category": {
+                        "enum": ["security", "compliance", "cost", "operational"]
+                    },
+                    "severity": {
+                        "enum": ["low", "medium", "high", "critical"]
+                    }
+                },
+                "additionalProperties": { "type": "any" }
+            }
+        },
+        "required": ["effect"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "effect": "audit",
+        "description": "Audit non-compliant resources",
+        "auditDetails": {
+            "category": "security",
+            "severity": "high"
+        }
+    }));
+
+    c.bench_function("validate_azure_policy_audit_effect", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_azure_policy_modify_effect - 1.17 µs/iter
+fn bench_azure_policy_modify_effect_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "effect": {
+                "const": "modify"
+            },
+            "description": {
+                "type": "string"
+            },
+            "modifyDetails": {
+                "type": "object",
+                "properties": {
+                    "roleDefinitionIds": {
+                        "type": "array",
+                        "items": { "type": "string" }
+                    },
+                    "operations": {
+                        "type": "array",
+                        "items": {
+                            "type": "object",
+                            "properties": {
+                                "operation": {
+                                    "enum": ["add", "replace", "remove"]
+                                },
+                                "field": {
+                                    "type": "string"
+                                },
+                                "value": {
+                                "type": "any"
+                            }
+                            },
+                            "required": ["operation", "field"],
+                            "additionalProperties": { "type": "any" }
+                        }
+                    }
+                },
+                "required": ["roleDefinitionIds", "operations"],
+                "additionalProperties": { "type": "any" }
+            }
+        },
+        "required": ["effect", "modifyDetails"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "effect": "modify",
+        "description": "Modify resources to ensure compliance",
+        "modifyDetails": {
+            "roleDefinitionIds": [
+                "/providers/Microsoft.Authorization/roleDefinitions/b24988ac-6180-42a0-ab88-20f7382dd24c"
+            ],
+            "operations": [
+                {
+                    "operation": "add",
+                    "field": "tags.environment",
+                    "value": "production"
+                }
+            ]
+        }
+    }));
+
+    c.bench_function("validate_azure_policy_modify_effect", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+// Observed: validate_azure_policy_complex_effect - 1.40 µs/iter
+fn bench_azure_policy_complex_effect_validation(c: &mut Criterion) {
+    let schema_json = json!({
+        "type": "object",
+        "properties": {
+            "effect": {
+                "enum": ["auditIfNotExists", "deployIfNotExists"]
+            },
+            "parameters": {
+                "type": "object",
+                "additionalProperties": { "type": "any" }
+            },
+            "existenceCondition": {
+                "type": "object",
+                "properties": {
+                    "field": { "type": "string" },
+                    "equals": { "type": "string" }
+                },
+                "required": ["field"],
+                "additionalProperties": { "type": "any" }
+            },
+            "deployment": {
+                "type": "object",
+                "properties": {
+                    "properties": {
+                        "type": "object",
+                        "properties": {
+                            "mode": {
+                                "enum": ["incremental", "complete"]
+                            },
+                            "template": {
+                                "type": "object",
+                                "additionalProperties": { "type": "any" }
+                            },
+                            "parameters": {
+                                "type": "object",
+                                "additionalProperties": { "type": "any" }
+                            }
+                        },
+                        "required": ["mode", "template"],
+                        "additionalProperties": { "type": "any" }
+                    }
+                },
+                "required": ["properties"],
+                "additionalProperties": { "type": "any" }
+            }
+        },
+        "required": ["effect"],
+        "additionalProperties": { "type": "any" }
+    });
+    let schema = Schema::from_serde_json_value(schema_json).unwrap();
+    let value = Value::from(json!({
+        "effect": "deployIfNotExists",
+        "parameters": {},
+        "existenceCondition": {
+            "field": "Microsoft.Security/complianceResults/resourceStatus",
+            "equals": "OffByPolicy"
+        },
+        "deployment": {
+            "properties": {
+                "mode": "incremental",
+                "template": {
+                    "$schema": "https://schema.management.azure.com/schemas/2015-01-01/deploymentTemplate.json#",
+                    "contentVersion": "1.0.0.0",
+                    "resources": []
+                },
+                "parameters": {}
+            }
+        }
+    }));
+
+    c.bench_function("validate_azure_policy_complex_effect", |b| {
+        b.iter(|| {
+            SchemaValidator::validate(&value, &schema).unwrap();
+        })
+    });
+}
+
+criterion_group!(
+    schema_validation_benches,
+    bench_string_validation,
+    bench_number_validation,
+    bench_array_validation,
+    bench_object_validation,
+    bench_complex_nested_validation,
+    bench_string_pattern_validation,
+    bench_enum_validation,
+    bench_boolean_validation,
+    bench_null_validation,
+    bench_large_array_validation,
+    bench_deeply_nested_object,
+    bench_mixed_type_array,
+    bench_additional_properties,
+    bench_array_constraints,
+    bench_multi_level_validation,
+    bench_azure_vm_resource_validation,
+    bench_azure_storage_resource_validation,
+    bench_azure_arm_template_validation,
+    bench_azure_policy_deny_effect_validation,
+    bench_azure_policy_audit_effect_validation,
+    bench_azure_policy_modify_effect_validation,
+    bench_azure_policy_complex_effect_validation
+);
+criterion_main!(schema_validation_benches);

bindings/c-nostd/CMakeLists.txt

@@ -7,7 +7,8 @@ include(FetchContent)
 FetchContent_Declare(
     Corrosion
     GIT_REPOSITORY https://github.com/corrosion-rs/corrosion.git
-    GIT_TAG v0.4 # Optionally specify a commit hash, version tag or branch here
+    # Use a tag that has a fix for https://github.com/corrosion-rs/corrosion/issues/590
+    GIT_TAG 6be991bb34c348dfb8344be22f3606288ea5c7fd
 )
 FetchContent_MakeAvailable(Corrosion)
 
@@ -31,6 +32,8 @@ corrosion_import_crate(
   # See regorus/opa_no_std
   FEATURES "custom_allocator,regorus/semver"
 
+  LOCKED
+  
   # Link statically
   CRATE_TYPES staticlib FLAGS --crate-type=staticlib
   )

bindings/c-nostd/main.c

@@ -1,92 +1,126 @@
 #include <stdio.h>
+#include <stdlib.h>
+#if defined(_WIN32)
+#include <malloc.h>
+#endif
 #include "regorus.h"
 
-
 // Regorus has been built for no_std and cannot access files.
-char*  file_to_string(const char* file) {
-    char * buffer = 0;
+char *file_to_string(const char *file)
+{
+    char *buffer = 0;
     long length;
-    FILE * f = fopen (file, "rb");
+    FILE *f = fopen(file, "rb");
 
     if (f)
     {
-	fseek (f, 0, SEEK_END);
-	length = ftell (f);
-	fseek (f, 0, SEEK_SET);
-	buffer = malloc (length + 1);
-	buffer[length] = '\0';
-	if (buffer)
-	{
-	    fread (buffer, 1, length, f);
-	}
-	fclose (f);
+        fseek(f, 0, SEEK_END);
+        length = ftell(f);
+        fseek(f, 0, SEEK_SET);
+        buffer = malloc(length + 1);
+        buffer[length] = '\0';
+        if (buffer)
+        {
+            fread(buffer, 1, length, f);
+        }
+        fclose(f);
     }
 
     return buffer;
 }
 
 // If regorus is built with custom-allocator, then provide implementation.
-uint8_t* regorus_aligned_alloc(size_t alignment, size_t size) {
-    return aligned_alloc(alignment, size);
+uint8_t *regorus_aligned_alloc(size_t alignment, size_t size)
+{
+    // Aligned allocations must respect platform quirks: Windows offers
+    // _aligned_malloc/_aligned_free, while macOS/Linux reject aligned_alloc
+    // calls when size is not a multiple of alignment, so we rely on
+    // posix_memalign for the no_std build.
+#if defined(_WIN32)
+    return (uint8_t *)_aligned_malloc(size, alignment);
+#else
+    void *ptr = NULL;
+    // posix_memalign requires alignment to be at least sizeof(void*)
+    // and a power of two; normalize here so small requests succeed.
+    if (alignment < sizeof(void *))
+    {
+        alignment = sizeof(void *);
+    }
+
+    if (posix_memalign(&ptr, alignment, size) != 0)
+    {
+        return NULL;
+    }
+    return (uint8_t *)ptr;
+#endif
 }
 
-void regorus_free(uint8_t* ptr) {
+void regorus_free(uint8_t *ptr)
+{
+#if defined(_WIN32)
+    _aligned_free(ptr);
+#else
     free(ptr);
+#endif
 }
 
-
-int main() {
+int main()
+{
     // Create engine.
-    RegorusEngine* engine = regorus_engine_new();
+    RegorusEngine *engine = regorus_engine_new();
     RegorusResult r;
-    char* buffer = NULL;
+    char *buffer = NULL;
+
+    // Turn on rego v0 since policy uses v0.
+    r = regorus_engine_set_rego_v0(engine, true);
+    if (r.status != Ok)
+        goto error;
 
     // Load policies.
     r = regorus_engine_add_policy(engine, "framework.rego", (buffer = file_to_string("../../../tests/aci/framework.rego")));
     free(buffer);
-    if (r.status != RegorusStatusOk)
-	goto error;
+    if (r.status != Ok)
+        goto error;
     printf("Loaded package %s\n", r.output);
     regorus_result_drop(r);
 
     r = regorus_engine_add_policy(engine, "api.rego", (buffer = file_to_string("../../../tests/aci/api.rego")));
     free(buffer);
-    if (r.status != RegorusStatusOk)
-	goto error;
+    if (r.status != Ok)
+        goto error;
     printf("Loaded package %s\n", r.output);
     regorus_result_drop(r);
 
     r = regorus_engine_add_policy(engine, "policy.rego", (buffer = file_to_string("../../../tests/aci/policy.rego")));
     free(buffer);
-    if (r.status != RegorusStatusOk)
-	goto error;
+    if (r.status != Ok)
+        goto error;
     printf("Loaded package %s\n", r.output);
     regorus_result_drop(r);
 
     // Add data
     r = regorus_engine_add_data_json(engine, (buffer = file_to_string("../../../tests/aci/data.json")));
     free(buffer);
-    if (r.status != RegorusStatusOk)
-	goto error;
+    if (r.status != Ok)
+        goto error;
     regorus_result_drop(r);
 
     // Set input
     r = regorus_engine_set_input_json(engine, (buffer = file_to_string("../../../tests/aci/input.json")));
     free(buffer);
-    if (r.status != RegorusStatusOk)
-	goto error;
+    if (r.status != Ok)
+        goto error;
     regorus_result_drop(r);
 
     // Eval rule.
     r = regorus_engine_eval_rule(engine, "data.framework.mount_overlay");
-    if (r.status != RegorusStatusOk)
-	goto error;
+    if (r.status != Ok)
+        goto error;
 
     // Print output
     printf("%s", r.output);
     regorus_result_drop(r);
 
-
     // Free the engine.
     regorus_engine_drop(engine);
 

bindings/c/CMakeLists.txt

@@ -7,11 +7,13 @@ include(FetchContent)
 FetchContent_Declare(
     Corrosion
     GIT_REPOSITORY https://github.com/corrosion-rs/corrosion.git
-    GIT_TAG v0.4 # Optionally specify a commit hash, version tag or branch here
+    # Use a tag that has a fix for https://github.com/corrosion-rs/corrosion/issues/590
+    GIT_TAG 6be991bb34c348dfb8344be22f3606288ea5c7fd
 )
 FetchContent_MakeAvailable(Corrosion)
 
 project("regorus-test")
+enable_testing()
 
 corrosion_import_crate(
   # Path to <regorus-source-folder>/bindings/ffi/Cargo.toml
@@ -24,6 +26,8 @@ corrosion_import_crate(
   # Select specific features in regorus.
   FEATURES "regorus/semver"
 
+  LOCKED
+  
   # Link statically
   CRATE_TYPES "cdylib"
   )
@@ -32,3 +36,10 @@ add_executable(regorus_test main.c)
 # Add path to <regorus-source-folder>/bindings/ffi
 target_include_directories(regorus_test PRIVATE "../ffi")
 target_link_libraries(regorus_test regorus_ffi)
+
+add_executable(regorus_rvm_test rvm_tests.c)
+target_include_directories(regorus_rvm_test PRIVATE "../ffi")
+target_link_libraries(regorus_rvm_test regorus_ffi)
+
+add_test(NAME regorus_c_engine COMMAND regorus_test)
+add_test(NAME regorus_c_rvm COMMAND regorus_rvm_test)

bindings/c/main.c

@@ -6,40 +6,59 @@ int main() {
     RegorusEngine* engine = regorus_engine_new();
     RegorusResult r;
 
+    // Turn on rego v0 since policy uses v0.
+    r = regorus_engine_set_rego_v0(engine, true);
+    if (r.status != Ok)
+	goto error;
+
+    // Configure the global pattern caches.
+    RegorusCacheConfig cache_config = { .regex = 256, .glob = 128 };
+    r = regorus_set_cache_config(cache_config);
+    if (r.status != Ok)
+	goto error;
+    regorus_result_drop(r);
+
+    // Raise the default col limit to 2000
+    RegorusPolicyLengthConfig len_config = { .max_col = 2000, .max_file_bytes = 1048576, .max_lines = 20000 };
+    r = regorus_engine_set_policy_length_config(engine, len_config);
+    if (r.status != Ok)
+	goto error;
+    regorus_result_drop(r);
+
     // Load policies.
     r = regorus_engine_add_policy_from_file(engine, "../../../tests/aci/framework.rego");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
     printf("Loaded package %s\n", r.output);
     regorus_result_drop(r);
 
     r = regorus_engine_add_policy_from_file(engine, "../../../tests/aci/api.rego");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
     printf("Loaded package %s\n", r.output);
     regorus_result_drop(r);
     
     r = regorus_engine_add_policy_from_file(engine, "../../../tests/aci/policy.rego");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
     printf("Loaded package %s\n", r.output);
     regorus_result_drop(r);
 
     // Add data
     r = regorus_engine_add_data_from_json_file(engine, "../../../tests/aci/data.json");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
     regorus_result_drop(r);
 
     // Set input
     r = regorus_engine_set_input_from_json_file(engine, "../../../tests/aci/input.json");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
     regorus_result_drop(r);
 
     // Eval rule.
     r = regorus_engine_eval_query(engine, "data.framework.mount_overlay");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
 
     // Print output
@@ -61,14 +80,14 @@ int main() {
 	);
 
     // Evaluate rule.
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
 
     r = regorus_engine_set_enable_coverage(engine, true);
     regorus_result_drop(r);
     
     r = regorus_engine_eval_query(engine, "data.test.message");
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
 
     // Print output
@@ -77,7 +96,7 @@ int main() {
 	
     // Print pretty coverage report.
     r = regorus_engine_get_coverage_report_pretty(engine);
-    if (r.status != RegorusStatusOk)
+    if (r.status != Ok)
 	goto error;
 
     printf("%s\n", r.output);

bindings/c/rvm_tests.c

@@ -0,0 +1,289 @@
+// Copyright (c) Microsoft Corporation.
+// Licensed under the MIT License.
+
+#include <stdio.h>
+#include <string.h>
+#include "regorus.h"
+
+static int assert_ok(RegorusResult r, const char* message) {
+    if (r.status != Ok) {
+        fprintf(stderr, "%s: %s\n", message, r.error_message ? r.error_message : "(no error)");
+        return 0;
+    }
+    return 1;
+}
+
+int main() {
+    RegorusResult result = {0};
+    bool result_valid = false;
+    RegorusProgram* program = NULL;
+    RegorusBuffer* buffer = NULL;
+    RegorusProgram* program2 = NULL;
+    RegorusRvm* vm = NULL;
+    RegorusProgram* host_program = NULL;
+    RegorusRvm* host_vm = NULL;
+    bool is_partial = false;
+    int exit_code = 1;
+
+    const char* data_json =
+        "{"
+        "  \"roles\": {"
+        "    \"alice\": [\"admin\", \"reader\"]"
+        "  }"
+        "}";
+    const char* input_json =
+        "{"
+        "  \"user\": \"alice\","
+        "  \"actions\": [\"read\"]"
+        "}";
+    const char* module_text =
+        "package demo\n"
+        "default allow = false\n"
+        "allow if {\n"
+        "  input.user == \"alice\"\n"
+        "  some role in data.roles[input.user]\n"
+        "  role == \"admin\"\n"
+        "  count(input.actions) > 0\n"
+        "}\n";
+
+    const char* host_data_json = "{}";
+    const char* host_input_json = "{\"account\":{\"id\":\"acct-1\",\"active\":true}}";
+    const char* host_module_text =
+        "package demo\n"
+        "import rego.v1\n"
+        "default allow := false\n"
+        "allow if {\n"
+        "  input.account.active == true\n"
+        "  details := __builtin_host_await(input.account.id, \"account\")\n"
+        "  details.tier == \"gold\"\n"
+        "}\n";
+
+    RegorusPolicyModule module;
+    module.id = "demo.rego";
+    module.content = module_text;
+
+    const char* entry_points[] = {"data.demo.allow"};
+    printf("Rego policy:\n%s\n", module_text);
+    printf("Compiling program from modules...\n");
+    result = regorus_program_compile_from_modules(
+        data_json,
+        &module,
+        1,
+        entry_points,
+        1
+    );
+    result_valid = true;
+    if (!assert_ok(result, "compile program")) {
+        goto Cleanup;
+    }
+    program = (RegorusProgram*)result.pointer_value;
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Generating assembly listing...\n");
+    result = regorus_program_generate_listing(program);
+    result_valid = true;
+    if (!assert_ok(result, "generate listing")) {
+        goto Cleanup;
+    }
+    printf("Assembly listing:\n%s\n", result.output ? result.output : "(null)");
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Serializing program...\n");
+    result = regorus_program_serialize_binary(program);
+    result_valid = true;
+    if (!assert_ok(result, "serialize program")) {
+        goto Cleanup;
+    }
+    buffer = (RegorusBuffer*)result.pointer_value;
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Deserializing program (%zu bytes)...\n", buffer->len);
+    result = regorus_program_deserialize_binary(
+        buffer->data,
+        buffer->len,
+        &is_partial
+    );
+    result_valid = true;
+    if (!assert_ok(result, "deserialize program")) {
+        goto Cleanup;
+    }
+
+    if (is_partial) {
+        fprintf(stderr, "deserialized program marked partial\n");
+        goto Cleanup;
+    }
+
+    program2 = (RegorusProgram*)result.pointer_value;
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Creating VM...\n");
+    vm = regorus_rvm_new();
+    if (!vm) {
+        fprintf(stderr, "failed to allocate VM\n");
+        goto Cleanup;
+    }
+
+    printf("Loading program into VM...\n");
+    result = regorus_rvm_load_program(vm, program2);
+    result_valid = true;
+    if (!assert_ok(result, "load program")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Setting data...\n");
+    result = regorus_rvm_set_data(vm, data_json);
+    result_valid = true;
+    if (!assert_ok(result, "set data")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Setting input...\n");
+    result = regorus_rvm_set_input(vm, input_json);
+    result_valid = true;
+    if (!assert_ok(result, "set input")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    printf("Executing entry point...\n");
+    result = regorus_rvm_execute(vm);
+    result_valid = true;
+    if (!assert_ok(result, "execute")) {
+        goto Cleanup;
+    }
+
+        printf("Execution result (data.demo.allow): %s\n",
+            result.output ? result.output : "(null)");
+        printf("Decision: user=alice action=read -> allow=%s\n",
+            result.output ? result.output : "(null)");
+    if (!result.output || strcmp(result.output, "true") != 0) {
+        fprintf(stderr, "unexpected result: %s\n", result.output);
+        goto Cleanup;
+    }
+
+    printf("\n--- HostAwait example (suspendable execution) ---\n");
+    RegorusPolicyModule host_module;
+    host_module.id = "host_await.rego";
+    host_module.content = host_module_text;
+
+    const char* host_entry_points[] = {"data.demo.allow"};
+    result = regorus_program_compile_from_modules(
+        host_data_json,
+        &host_module,
+        1,
+        host_entry_points,
+        1
+    );
+    result_valid = true;
+    if (!assert_ok(result, "compile host await program")) {
+        goto Cleanup;
+    }
+    host_program = (RegorusProgram*)result.pointer_value;
+    regorus_result_drop(result);
+    result_valid = false;
+
+    host_vm = regorus_rvm_new();
+    if (!host_vm) {
+        fprintf(stderr, "failed to allocate host await VM\n");
+        goto Cleanup;
+    }
+
+    result = regorus_rvm_set_execution_mode(host_vm, 1);
+    result_valid = true;
+    if (!assert_ok(result, "set execution mode")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    result = regorus_rvm_load_program(host_vm, host_program);
+    result_valid = true;
+    if (!assert_ok(result, "load host await program")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    result = regorus_rvm_set_data(host_vm, host_data_json);
+    result_valid = true;
+    if (!assert_ok(result, "set host data")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    result = regorus_rvm_set_input(host_vm, host_input_json);
+    result_valid = true;
+    if (!assert_ok(result, "set host input")) {
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    result = regorus_rvm_execute(host_vm);
+    result_valid = true;
+    if (!assert_ok(result, "execute host await")) {
+        goto Cleanup;
+    }
+    printf("HostAwait initial result: %s\n", result.output ? result.output : "(null)");
+    regorus_result_drop(result);
+    result_valid = false;
+
+    result = regorus_rvm_get_execution_state(host_vm);
+    result_valid = true;
+    if (!assert_ok(result, "get execution state")) {
+        goto Cleanup;
+    }
+    printf("Execution state: %s\n", result.output ? result.output : "(null)");
+    regorus_result_drop(result);
+    result_valid = false;
+
+    result = regorus_rvm_resume(host_vm, "{\"tier\":\"gold\"}", true);
+    result_valid = true;
+    if (!assert_ok(result, "resume host await")) {
+        goto Cleanup;
+    }
+    printf("HostAwait resumed result: %s\n", result.output ? result.output : "(null)");
+
+    if (!result.output || strcmp(result.output, "true") != 0) {
+        fprintf(stderr, "unexpected host await result\n");
+        goto Cleanup;
+    }
+    regorus_result_drop(result);
+    result_valid = false;
+
+    exit_code = 0;
+
+Cleanup:
+    if (result_valid) {
+        regorus_result_drop(result);
+    }
+    if (host_vm) {
+        regorus_rvm_drop(host_vm);
+    }
+    if (host_program) {
+        regorus_program_drop(host_program);
+    }
+    if (vm) {
+        regorus_rvm_drop(vm);
+    }
+    if (program2) {
+        regorus_program_drop(program2);
+    }
+    if (buffer) {
+        regorus_buffer_drop(buffer);
+    }
+    if (program) {
+        regorus_program_drop(program);
+    }
+    return exit_code;
+}

bindings/cpp/CMakeLists.txt

@@ -7,12 +7,16 @@ include(FetchContent)
 FetchContent_Declare(
     Corrosion
     GIT_REPOSITORY https://github.com/corrosion-rs/corrosion.git
-    GIT_TAG v0.4 # Optionally specify a commit hash, version tag or branch here
+    # Use a tag that has a fix for https://github.com/corrosion-rs/corrosion/issues/590
+    GIT_TAG 6be991bb34c348dfb8344be22f3606288ea5c7fd
 )
 FetchContent_MakeAvailable(Corrosion)
 
 project("regorus-test")
 set(CMAKE_CXX_STANDARD 17)
+enable_testing()
+
+# installable ffi target
 
 corrosion_import_crate(
   # Path to <regorus-source-folder>/bindings/ffi/Cargo.toml
@@ -25,10 +29,64 @@ corrosion_import_crate(
   # Select specific features in regorus.
   FEATURES "regorus/semver"
 
+  LOCKED
+  
   # Link statically
   CRATE_TYPES "cdylib")
 
+include(GNUInstallDirs)
+
+set(regorus_ffi_INCLUDEDIR ${CMAKE_INSTALL_INCLUDEDIR}/regorus_ffi)
+set(regorus_ffi_CONFIGDIR ${CMAKE_INSTALL_LIBDIR}/cmake/regorus_ffi)
+set(regorus_ffi_LIBDIR ${CMAKE_INSTALL_LIBDIR})
+set(regorus_ffi_BINDIR ${CMAKE_INSTALL_BINDIR})
+
+add_library(regorus_ffi::regorus_ffi ALIAS regorus_ffi)
+corrosion_install(TARGETS regorus_ffi EXPORT regorus_ffi_targets LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR})
+
+target_include_directories(regorus_ffi
+INTERFACE
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}>
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/../ffi>
+    $<INSTALL_INTERFACE:${regorus_ffi_INCLUDEDIR}>
+)
+
+set(regorus_ffi_HEADER_FILES
+  regorus.hpp
+  ../ffi/regorus.ffi.hpp
+)
+
+install(FILES ${regorus_ffi_HEADER_FILES}
+  DESTINATION ${regorus_ffi_INCLUDEDIR}
+  COMPONENT Devel
+)
+
+install(EXPORT regorus_ffi_targets
+  FILE regorus_ffi_targets.cmake
+  NAMESPACE regorus_ffi::
+  DESTINATION ${regorus_ffi_CONFIGDIR}
+)
+
+include(CMakePackageConfigHelpers)
+
+configure_package_config_file(${CMAKE_CURRENT_LIST_DIR}/regorus_ffiConfig.cmake.in
+  ${CMAKE_CURRENT_BINARY_DIR}/regorus_ffiConfig.cmake
+  INSTALL_DESTINATION ${regorus_ffi_CONFIGDIR}
+)
+
+install(FILES
+  ${CMAKE_CURRENT_BINARY_DIR}/regorus_ffiConfig.cmake
+  ${CMAKE_CURRENT_BINARY_DIR}/corrosion/regorus_ffi_targetsCorrosion.cmake
+  DESTINATION ${regorus_ffi_CONFIGDIR}
+)
+
+# test binary
+
 add_executable(regorus_test main.cpp)
-# Add path to <regorus-source-folder>/bindings/ffi
-target_include_directories(regorus_test PRIVATE "../ffi")
-target_link_libraries(regorus_test regorus_ffi)
+target_link_libraries(regorus_test regorus_ffi::regorus_ffi)
+
+add_executable(regorus_rvm_test rvm_tests.cpp)
+target_link_libraries(regorus_rvm_test regorus_ffi::regorus_ffi)
+
+add_test(NAME regorus_cpp_engine COMMAND regorus_test)
+add_test(NAME regorus_cpp_rvm COMMAND regorus_rvm_test)

bindings/cpp/main.cpp

@@ -6,7 +6,19 @@ void example()
     // Create engine
     regorus::Engine engine;
 
+    // Configure the global pattern caches.
+    RegorusCacheConfig cache_config = { 256, 128 };
+    regorus::set_cache_config(cache_config);
+
+    engine.set_rego_v0(true);
     engine.set_enable_coverage(true);
+
+    RegorusPolicyLengthConfig len_config;
+    // Raise the default col limit to 2000
+    len_config.max_col = 2000;
+    len_config.max_file_bytes = 1048576;
+    len_config.max_lines = 20000;
+    engine.set_policy_length_config(len_config);
     
     // Add policies.
     engine.add_policy("objects.rego",R"(package objects
@@ -83,6 +95,7 @@ int main() {
 
     // Create engine.
     regorus::Engine engine;
+    engine.set_rego_v0(true);
 
 
     // Load policies.

bindings/cpp/regorus.hpp

@@ -1,6 +1,8 @@
 #ifndef REGORUS_WRAPPER_HPP
 #define REGORUS_WRAPPER_HPP
 
+#include <cstddef>
+#include <cstdint>
 #include <memory>
 #include <variant>
 
@@ -8,11 +10,14 @@
 
 namespace regorus {
 
-    class Result {
-    public:
+	class Buffer;
+	class Program;
 
-	operator bool() const { return result.status == RegorusStatus::RegorusStatusOk; }
-	bool operator !() const { return result.status != RegorusStatus::RegorusStatusOk; }
+	class Result {
+	public:
+
+	operator bool() const { return result.status == RegorusStatus::Ok; }
+	bool operator !() const { return result.status != RegorusStatus::Ok; }
 
 	const char* output() const {
 	    if (*this && result.output) {
@@ -30,18 +35,39 @@ namespace regorus {
 	    }
 	}
 
+	void* pointer() const {
+	    return result.pointer_value;
+	}
+
+	Program program() const;
+	Buffer buffer() const;
+
+	Result(RegorusResult r) : result(r) {}
+	Result(Result&& other) noexcept : result(other.result) {
+	    other.result.output = nullptr;
+	    other.result.error_message = nullptr;
+	    other.result.pointer_value = nullptr;
+	}
+	Result& operator=(Result&& other) noexcept {
+	    if (this != &other) {
+		regorus_result_drop(result);
+		result = other.result;
+		other.result.output = nullptr;
+		other.result.error_message = nullptr;
+		other.result.pointer_value = nullptr;
+	    }
+	    return *this;
+	}
+
 	~Result() {
 	    regorus_result_drop(result);
 	}
 	
     private:
-	friend class Engine;
 	RegorusResult result;
 
-	Result(RegorusResult r) : result(r) {}	
     private:
 	Result(const Result&) = delete;
-	Result(Result&&) = delete;
 	Result& operator=(const Result&) = delete;
 
     };
@@ -54,6 +80,9 @@ namespace regorus {
 	    return std::unique_ptr<Engine>(new Engine(regorus_engine_clone(engine)));
 	}
 
+	Result set_rego_v0(bool enable) {
+		return Result(regorus_engine_set_rego_v0(engine, enable));
+	}
 
 	Result add_policy(const char* path, const char* policy) {
 	    return Result(regorus_engine_add_policy(engine, path, policy));
@@ -102,10 +131,22 @@ namespace regorus {
 	Result get_coverage_report_pretty() {
             return Result(regorus_engine_get_coverage_report_pretty(engine));
 	}
-	
+
+	Result set_policy_length_config(RegorusPolicyLengthConfig config) {
+	    return Result(regorus_engine_set_policy_length_config(engine, config));
+	}
+
+	Result clear_policy_length_config() {
+	    return Result(regorus_engine_clear_policy_length_config(engine));
+	}
+
 	~Engine() {
 	    regorus_engine_drop(engine);
 	}
+
+	RegorusEngine* raw() const {
+	    return engine;
+	}
 	    
 	
     private:
@@ -116,6 +157,255 @@ namespace regorus {
 	Engine(Engine&&) = delete;
 	Engine& operator=(const Engine&) = delete;
     };
+
+    inline Result set_cache_config(RegorusCacheConfig config) {
+	    return Result(regorus_set_cache_config(config));
+    }
+
+    inline Result clear_cache() {
+	    return Result(regorus_clear_cache());
+    }
+
+    class CompiledPolicy {
+    public:
+	explicit CompiledPolicy(RegorusCompiledPolicy* p) : policy(p) {}
+
+	Result eval_with_input(const char* input_json) {
+	    return Result(regorus_compiled_policy_eval_with_input(policy, input_json));
+	}
+
+	Result get_policy_info() {
+	    return Result(regorus_compiled_policy_get_policy_info(policy));
+	}
+
+	RegorusCompiledPolicy* raw() const {
+	    return policy;
+	}
+
+	~CompiledPolicy() {
+	    if (policy) {
+		regorus_compiled_policy_drop(policy);
+	    }
+	}
+
+    private:
+	RegorusCompiledPolicy* policy;
+	CompiledPolicy(const CompiledPolicy&) = delete;
+	CompiledPolicy(CompiledPolicy&&) = delete;
+	CompiledPolicy& operator=(const CompiledPolicy&) = delete;
+    };
+
+    class Buffer {
+    public:
+	Buffer() : buffer(nullptr) {}
+	explicit Buffer(RegorusBuffer* b) : buffer(b) {}
+
+	const std::uint8_t* data() const {
+	    return buffer ? buffer->data : nullptr;
+	}
+
+	size_t size() const {
+	    return buffer ? buffer->len : 0;
+	}
+
+	RegorusBuffer* raw() const {
+	    return buffer;
+	}
+
+	~Buffer() {
+	    if (buffer) {
+		regorus_buffer_drop(buffer);
+	    }
+	}
+
+    private:
+	RegorusBuffer* buffer;
+	Buffer(const Buffer&) = delete;
+	Buffer(Buffer&&) = delete;
+	Buffer& operator=(const Buffer&) = delete;
+    };
+
+    class Program {
+    public:
+	Program() : program(regorus_program_new()) {}
+	explicit Program(RegorusProgram* p) : program(p) {}
+
+	static Result compile_from_policy(
+	    RegorusCompiledPolicy* compiled_policy,
+	    const char* const* entry_points,
+	    size_t entry_points_len
+	) {
+	    return Result(regorus_program_compile_from_policy(
+		compiled_policy,
+		entry_points,
+		entry_points_len
+	    ));
+	}
+
+	static Result compile_from_modules(
+	    const char* data_json,
+	    const RegorusPolicyModule* modules,
+	    size_t modules_len,
+	    const char* const* entry_points,
+	    size_t entry_points_len
+	) {
+	    return Result(regorus_program_compile_from_modules(
+		data_json,
+		modules,
+		modules_len,
+		entry_points,
+		entry_points_len
+	    ));
+	}
+
+	static Result compile_from_engine(
+	    RegorusEngine* engine,
+	    const char* const* entry_points,
+	    size_t entry_points_len
+	) {
+	    return Result(regorus_engine_compile_program_with_entrypoints(
+		engine,
+		entry_points,
+		entry_points_len
+	    ));
+	}
+
+	Result serialize_binary() const {
+	    return Result(regorus_program_serialize_binary(program));
+	}
+
+	static Result deserialize_binary(
+	    const std::uint8_t* data,
+	    size_t len,
+	    bool* is_partial
+	) {
+	    return Result(regorus_program_deserialize_binary(data, len, is_partial));
+	}
+
+	Result generate_listing() const {
+	    return Result(regorus_program_generate_listing(program));
+	}
+
+	Result generate_tabular_listing() const {
+	    return Result(regorus_program_generate_tabular_listing(program));
+	}
+
+	RegorusProgram* raw() const {
+	    return program;
+	}
+
+	~Program() {
+	    if (program) {
+		regorus_program_drop(program);
+	    }
+	}
+
+    private:
+	RegorusProgram* program;
+	Program(const Program&) = delete;
+	Program(Program&&) = delete;
+	Program& operator=(const Program&) = delete;
+    };
+
+	inline Program Result::program() const {
+	return Program(reinterpret_cast<RegorusProgram*>(result.pointer_value));
+	}
+
+	inline Buffer Result::buffer() const {
+	return Buffer(reinterpret_cast<RegorusBuffer*>(result.pointer_value));
+	}
+
+    class Rvm {
+    public:
+	Rvm() : vm(regorus_rvm_new()) {}
+	explicit Rvm(RegorusRvm* v) : vm(v) {}
+
+	static Result create_with_policy(RegorusCompiledPolicy* compiled_policy) {
+	    return Result(regorus_rvm_new_with_policy(compiled_policy));
+	}
+
+	Result load_program(const Program& program) {
+	    return Result(regorus_rvm_load_program(vm, program.raw()));
+	}
+
+	Result set_data(const char* data_json) {
+	    return Result(regorus_rvm_set_data(vm, data_json));
+	}
+
+	Result set_input(const char* input_json) {
+	    return Result(regorus_rvm_set_input(vm, input_json));
+	}
+
+	Result set_max_instructions(size_t max_instructions) {
+	    return Result(regorus_rvm_set_max_instructions(vm, max_instructions));
+	}
+
+	Result set_strict_builtin_errors(bool strict) {
+	    return Result(regorus_rvm_set_strict_builtin_errors(vm, strict));
+	}
+
+	Result set_execution_mode(std::uint8_t mode) {
+	    return Result(regorus_rvm_set_execution_mode(vm, mode));
+	}
+
+	Result set_step_mode(bool enabled) {
+	    return Result(regorus_rvm_set_step_mode(vm, enabled));
+	}
+
+	Result set_execution_timer_config(bool has_config, RegorusExecutionTimerConfig config) {
+	    return Result(regorus_rvm_set_execution_timer_config(vm, has_config, config));
+	}
+
+	Result execute() {
+	    return Result(regorus_rvm_execute(vm));
+	}
+
+	Result execute_entry_point_by_name(const char* entry_point) {
+	    return Result(regorus_rvm_execute_entry_point_by_name(vm, entry_point));
+	}
+
+	Result execute_entry_point_by_index(size_t index) {
+	    return Result(regorus_rvm_execute_entry_point_by_index(vm, index));
+	}
+
+	Result resume(const char* resume_value_json, bool has_value) {
+	    return Result(regorus_rvm_resume(vm, resume_value_json, has_value));
+	}
+
+	Result get_execution_state() {
+	    return Result(regorus_rvm_get_execution_state(vm));
+	}
+
+	RegorusRvm* raw() const {
+	    return vm;
+	}
+
+	~Rvm() {
+	    if (vm) {
+		regorus_rvm_drop(vm);
+	    }
+	}
+
+    private:
+	RegorusRvm* vm;
+	Rvm(const Rvm&) = delete;
+	Rvm(Rvm&&) = delete;
+	Rvm& operator=(const Rvm&) = delete;
+    };
+
+    inline Result compile_policy_with_entrypoint(
+	const char* data_json,
+	const RegorusPolicyModule* modules,
+	size_t modules_len,
+	const char* entry_point
+    ) {
+	return Result(regorus_compile_policy_with_entrypoint(
+	    data_json,
+	    modules,
+	    modules_len,
+	    entry_point
+	));
+    }
 }
 
 #endif // REGORUS_WRAPPER_HPP

bindings/cpp/regorus_ffiConfig.cmake.in

@@ -0,0 +1,3 @@
+@PACKAGE_INIT@
+include("${CMAKE_CURRENT_LIST_DIR}/regorus_ffi_targets.cmake")
+include("${CMAKE_CURRENT_LIST_DIR}/regorus_ffi_targetsCorrosion.cmake")

bindings/cpp/rvm_tests.cpp

@@ -0,0 +1,261 @@
+// Copyright (c) Microsoft Corporation.
+// Licensed under the MIT License.
+
+#include <iostream>
+#include <string>
+#include "regorus.hpp"
+
+int main() {
+    const char* data_json =
+        "{"
+        "  \"roles\": {"
+        "    \"alice\": [\"admin\", \"reader\"]"
+        "  }"
+        "}";
+    const char* input_json =
+        "{"
+        "  \"user\": \"alice\","
+        "  \"actions\": [\"read\"]"
+        "}";
+    const char* module_text =
+        "package demo\n"
+        "default allow = false\n"
+        "allow if {\n"
+        "  input.user == \"alice\"\n"
+        "  some role in data.roles[input.user]\n"
+        "  role == \"admin\"\n"
+        "  count(input.actions) > 0\n"
+        "}\n";
+
+    const char* host_data_json = "{}";
+    const char* host_input_json = "{\"account\":{\"id\":\"acct-1\",\"active\":true}}";
+    const char* host_module_text =
+        "package demo\n"
+        "import rego.v1\n"
+        "default allow := false\n"
+        "allow if {\n"
+        "  input.account.active == true\n"
+        "  details := __builtin_host_await(input.account.id, \"account\")\n"
+        "  details.tier == \"gold\"\n"
+        "}\n";
+
+    RegorusPolicyModule module;
+    module.id = "demo.rego";
+    module.content = module_text;
+
+    const char* entry_points[] = {"data.demo.allow"};
+    std::cout << "Rego policy:\n" << module_text << std::endl;
+    std::cout << "Compiling program from modules..." << std::endl;
+    auto program_result = regorus::Program::compile_from_modules(
+        data_json,
+        &module,
+        1,
+        entry_points,
+        1
+    );
+    if (!program_result) {
+        std::cerr << "compile program (modules): " << program_result.error() << std::endl;
+        return 1;
+    }
+
+    regorus::Program program = program_result.program();
+
+    std::cout << "Generating assembly listing..." << std::endl;
+    auto listing_result = program.generate_listing();
+    if (!listing_result) {
+        std::cerr << "generate listing: " << listing_result.error() << std::endl;
+        return 1;
+    }
+    std::cout << "Assembly listing:\n" << listing_result.output() << std::endl;
+
+    std::cout << "Serializing program..." << std::endl;
+    auto serialize_result = program.serialize_binary();
+    if (!serialize_result) {
+        std::cerr << "serialize program: " << serialize_result.error() << std::endl;
+        return 1;
+    }
+
+    regorus::Buffer buffer(reinterpret_cast<RegorusBuffer*>(serialize_result.pointer()));
+    bool is_partial = false;
+    std::cout << "Deserializing program (" << buffer.size() << " bytes)..." << std::endl;
+    auto deserialize_result = regorus::Program::deserialize_binary(
+        buffer.data(),
+        buffer.size(),
+        &is_partial
+    );
+    if (!deserialize_result) {
+        std::cerr << "deserialize program: " << deserialize_result.error() << std::endl;
+        return 1;
+    }
+
+    if (is_partial) {
+        std::cerr << "deserialized program marked partial" << std::endl;
+        return 1;
+    }
+
+    regorus::Program program2 = deserialize_result.program();
+
+    {
+        std::cout << "Creating VM..." << std::endl;
+        regorus::Rvm vm;
+        auto load_result = vm.load_program(program2);
+        if (!load_result) {
+            std::cerr << "load program: " << load_result.error() << std::endl;
+            return 1;
+        }
+
+        std::cout << "Setting data..." << std::endl;
+        auto data_result = vm.set_data(data_json);
+        if (!data_result) {
+            std::cerr << "set data: " << data_result.error() << std::endl;
+            return 1;
+        }
+
+        std::cout << "Setting input..." << std::endl;
+        auto input_result = vm.set_input(input_json);
+        if (!input_result) {
+            std::cerr << "set input: " << input_result.error() << std::endl;
+            return 1;
+        }
+
+        std::cout << "Executing entry point..." << std::endl;
+        auto exec_result = vm.execute();
+        if (!exec_result) {
+            std::cerr << "execute: " << exec_result.error() << std::endl;
+            return 1;
+        }
+
+        std::cout << "Execution result (data.demo.allow): " << exec_result.output() << std::endl;
+        std::cout << "Decision: user=alice action=read -> allow=" << exec_result.output() << std::endl;
+        if (std::string(exec_result.output()) != "true") {
+            std::cerr << "unexpected result: " << exec_result.output() << std::endl;
+            return 1;
+        }
+    }
+
+    regorus::Engine engine;
+    std::cout << "Compiling program from engine..." << std::endl;
+    auto add_policy_result = engine.add_policy("demo.rego", module_text);
+    if (!add_policy_result) {
+        std::cerr << "engine add policy: " << add_policy_result.error() << std::endl;
+        return 1;
+    }
+
+    auto engine_program_result = regorus::Program::compile_from_engine(
+        engine.raw(),
+        entry_points,
+        1
+    );
+    if (!engine_program_result) {
+        std::cerr << "compile program (engine): " << engine_program_result.error() << std::endl;
+        return 1;
+    }
+
+    regorus::Program engine_program = engine_program_result.program();
+
+    regorus::Rvm engine_vm;
+    auto engine_load_result = engine_vm.load_program(engine_program);
+    if (!engine_load_result) {
+        std::cerr << "engine load program: " << engine_load_result.error() << std::endl;
+        return 1;
+    }
+
+    std::cout << "Setting engine data..." << std::endl;
+    auto engine_data_result = engine_vm.set_data(data_json);
+    if (!engine_data_result) {
+        std::cerr << "engine set data: " << engine_data_result.error() << std::endl;
+        return 1;
+    }
+
+    std::cout << "Setting engine input..." << std::endl;
+    auto engine_input_result = engine_vm.set_input(input_json);
+    if (!engine_input_result) {
+        std::cerr << "engine set input: " << engine_input_result.error() << std::endl;
+        return 1;
+    }
+
+    std::cout << "Executing engine entry point..." << std::endl;
+    auto engine_exec_result = engine_vm.execute();
+    if (!engine_exec_result) {
+        std::cerr << "engine execute: " << engine_exec_result.error() << std::endl;
+        return 1;
+    }
+
+    std::cout << "Engine execution result (data.demo.allow): " << engine_exec_result.output() << std::endl;
+    std::cout << "Decision: user=alice action=read -> allow=" << engine_exec_result.output() << std::endl;
+    if (std::string(engine_exec_result.output()) != "true") {
+        std::cerr << "unexpected engine result: " << engine_exec_result.output() << std::endl;
+        return 1;
+    }
+
+    std::cout << "\n--- HostAwait example (suspendable execution) ---" << std::endl;
+    RegorusPolicyModule host_module;
+    host_module.id = "host_await.rego";
+    host_module.content = host_module_text;
+    const char* host_entry_points[] = {"data.demo.allow"};
+
+    auto host_program_result = regorus::Program::compile_from_modules(
+        host_data_json,
+        &host_module,
+        1,
+        host_entry_points,
+        1
+    );
+    if (!host_program_result) {
+        std::cerr << "compile host await program: " << host_program_result.error() << std::endl;
+        return 1;
+    }
+
+    regorus::Program host_program = host_program_result.program();
+    regorus::Rvm host_vm;
+    auto host_mode_result = host_vm.set_execution_mode(1);
+    if (!host_mode_result) {
+        std::cerr << "set execution mode: " << host_mode_result.error() << std::endl;
+        return 1;
+    }
+
+    auto host_load_result = host_vm.load_program(host_program);
+    if (!host_load_result) {
+        std::cerr << "load host await program: " << host_load_result.error() << std::endl;
+        return 1;
+    }
+
+    auto host_data_result = host_vm.set_data(host_data_json);
+    if (!host_data_result) {
+        std::cerr << "set host data: " << host_data_result.error() << std::endl;
+        return 1;
+    }
+
+    auto host_input_result = host_vm.set_input(host_input_json);
+    if (!host_input_result) {
+        std::cerr << "set host input: " << host_input_result.error() << std::endl;
+        return 1;
+    }
+
+    auto host_exec_result = host_vm.execute();
+    if (!host_exec_result) {
+        std::cerr << "execute host await: " << host_exec_result.error() << std::endl;
+        return 1;
+    }
+    std::cout << "HostAwait initial result: " << host_exec_result.output() << std::endl;
+
+    auto host_state_result = host_vm.get_execution_state();
+    if (!host_state_result) {
+        std::cerr << "get execution state: " << host_state_result.error() << std::endl;
+        return 1;
+    }
+    std::cout << "Execution state: " << host_state_result.output() << std::endl;
+
+    auto host_resume_result = host_vm.resume("{\"tier\":\"gold\"}", true);
+    if (!host_resume_result) {
+        std::cerr << "resume host await: " << host_resume_result.error() << std::endl;
+        return 1;
+    }
+    std::cout << "HostAwait resumed result: " << host_resume_result.output() << std::endl;
+    if (std::string(host_resume_result.output()) != "true") {
+        std::cerr << "unexpected host await result: " << host_resume_result.output() << std::endl;
+        return 1;
+    }
+
+    return 0;
+}

bindings/csharp/API.md

@@ -0,0 +1,421 @@
+# Regorus C# API Documentation
+
+This document describes the C# API for Regorus, focusing on the compiled policy approach for high-performance policy evaluation.
+
+## Overview
+
+The Regorus C# bindings provide a modern, thread-safe API for compiling and evaluating Open Policy Agent (OPA) Rego policies. The API is designed around pre-compiled policies that can be evaluated efficiently multiple times with different inputs.
+
+## Architecture
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│                    CompiledPolicy Workflow                      │
+└─────────────────────────────────────────────────────────────────┘
+
+┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
+│  Policy Modules │    │   Target/Schema  │    │   Static Data   │
+│  (.rego files)  │    │    Registries    │    │     (JSON)      │
+└─────────┬───────┘    └────────┬─────────┘    └─────────┬───────┘
+          │                     │                        │
+          └─────────────────────┼────────────────────────┘
+                                │
+                                ▼
+                    ┌─────────────────────────┐
+                    │        Compile          │
+                    │  ┌─────────────────────┐│
+                    │  │ Parse & Analyze     ││
+                    │  │ Infer Resource Types││
+                    │  │ Build AST & Rules   ││
+                    │  │ Target Integration  ││
+                    │  └─────────────────────┘│
+                    └─────────────┬───────────┘
+                                  │
+                                  ▼
+                      ┌─────────────────────────┐
+                      │    CompiledPolicy       │
+                      │ ┌─────────────────────┐ │
+                      │ │ AST & Rules         │ │
+                      │ │ Target Info         │ │
+                      │ │ Resource Types      │ │
+                      │ │ Function Table      │ │
+                      │ │ Compiled Modules    │ │
+                      │ └─────────────────────┘ │
+                      └─────────────┬───────────┘
+                                    │
+                                    ▼
+                          ┌─────────────────────┐
+                          │    Service Cache    │
+                          │ (Policy Framework,  │
+                          │  MS Graph, etc.)    │
+                          │ ┌─────────────────┐ │
+                          │ │ CompiledPolicy  │ │ ◄─── Same LOCK-FREE policy
+                          │ │    (cached)     │ │      instance shared across
+                          │ └─────────────────┘ │      all threads
+                          └─────────┬───────────┘
+                                    │
+                            ┌───────┼───────┬───────┐
+                            │       │       │       │
+                            ▼       ▼       ▼       ▼
+                    ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
+                    │  Thread 1   │ │  Thread 2   │ │  Thread N   │
+                    │             │ │             │ │             │
+                    │ input1 ────▶│ │ input2 ────▶│ │ inputN ────▶│
+                    │ ◄─── result │ │ ◄─── result │ │ ◄─── result │
+                    └─────────────┘ └─────────────┘ └─────────────┘
+
+┌─────────────────────────────────────────────────────────────────┐
+│                        Key Benefits                             │
+├─────────────────────────────────────────────────────────────────┤
+│ ✓ Compile Once, Evaluate Many     ✓ Lock-Free Concurrent Eval   │
+│ ✓ No Re-parsing Overhead          ✓ Reference Counting Safety   │
+│ ✓ Reduced GC Pressure             ✓ Proper Resource Management  │
+│ ✓ Cache-Friendly Design           ✓ Target System Integration   │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+## Key Features
+
+- **Pre-compiled Policies**: Compile once, evaluate many times for optimal performance
+- **Target System Support**: Built-in support for Azure Policy targets with resource type inference
+- **Thread Safety**: All operations are thread-safe without external synchronization
+- **Registry Management**: Centralized management of targets and schemas
+- **Policy Introspection**: Rich metadata about compiled policies
+
+## Core Classes
+
+### CompiledPolicy
+
+The `CompiledPolicy` class represents a pre-compiled Rego policy that can be evaluated efficiently.
+
+```csharp
+public sealed class CompiledPolicy : IDisposable
+{
+    // Evaluate the policy with input data
+    public string? EvalWithInput(string inputJson);
+    
+    // Get comprehensive policy metadata
+    public PolicyInfo GetPolicyInfo();
+    
+    // Dispose of unmanaged resources
+    public void Dispose();
+}
+```
+
+**Thread Safety**: All methods are thread-safe. Multiple threads can call `EvalWithInput()` concurrently, and `Dispose()` will safely wait for active evaluations to complete.
+
+### Compiler
+
+The `Compiler` class provides static methods for compiling policies.
+
+```csharp
+public static class Compiler
+{
+    // Compile a policy with a specific entrypoint rule
+    public static CompiledPolicy CompilePolicyWithEntrypoint(
+        string dataJson, 
+        IEnumerable<PolicyModule> modules, 
+        string entryPointRule);
+    
+    // Compile a target-aware policy (requires azure_policy feature)
+    public static CompiledPolicy CompilePolicyForTarget(
+        string dataJson, 
+        IEnumerable<PolicyModule> modules);
+}
+```
+
+### PolicyModule
+
+Represents a single policy module to be compiled. Each PolicyModule corresponds to a Rego file (.rego), and each Rego file defines a Rego package using the `package` declaration at the top of the file.
+
+```csharp
+public struct PolicyModule
+{
+    public string Id { get; set; }
+    public string Content { get; set; }
+    
+    public PolicyModule(string id, string content);
+}
+```
+
+**Properties:**
+- `Id`: A unique identifier for the module, typically the filename (e.g., "policy.rego", "rules/storage.rego")
+- `Content`: The complete Rego policy content, including the `package` declaration and all rules
+
+**Example:**
+```csharp
+var module = new PolicyModule("storage-policy.rego", @"
+    package azure.storage
+    import rego.v1
+    
+    default allow := false
+    allow if input.type == ""Microsoft.Storage/storageAccounts""
+");
+```
+
+### PolicyInfo
+
+Provides comprehensive metadata about a compiled policy.
+
+```csharp
+public class PolicyInfo
+{
+    // List of module identifiers
+    public List<string> ModuleIds { get; set; }
+    
+    // Target name (for target-aware policies)
+    public string? TargetName { get; set; }
+    
+    // Resource types this policy can evaluate
+    public List<string> ApplicableResourceTypes { get; set; }
+    
+    // Primary rule/entrypoint
+    public string EntrypointRule { get; set; }
+    
+    // Effect rule (for target-aware policies)
+    public string? EffectRule { get; set; }
+    
+    // Policy parameters
+    public List<PolicyParameters> Parameters { get; set; }
+}
+```
+
+## Registry Classes
+
+### TargetRegistry
+
+Manages target definitions for Azure Policy-style evaluations.
+
+```csharp
+public static class TargetRegistry
+{
+    // Register a target from JSON
+    public static void RegisterFromJson(string targetJson);
+    
+    // Check if a target exists
+    public static bool Contains(string name);
+    
+    // List all registered targets
+    public static string ListNames();
+    
+    // Remove a target
+    public static bool Remove(string name);
+    
+    // Clear all targets
+    public static void Clear();
+    
+    // Get count of registered targets
+    public static int Count { get; }
+    
+    // Check if registry is empty
+    public static bool IsEmpty { get; }
+}
+```
+
+### SchemaRegistry
+
+Manages schema definitions for validation.
+
+```csharp
+public static class SchemaRegistry
+{
+    // Register resource schemas
+    public static void RegisterResourceSchema(string name, string schemaJson);
+    public static bool ContainsResourceSchema(string name);
+    public static string ListResourceSchemas();
+    
+    // Register effect schemas
+    public static void RegisterEffectSchema(string name, string schemaJson);
+    public static bool ContainsEffectSchema(string name);
+    public static string ListEffectSchemas();
+    
+    // Clear methods
+    public static void ClearResourceSchemas();
+    public static void ClearEffectSchemas();
+}
+```
+
+## Usage Examples
+
+### Basic Policy Compilation and Evaluation
+
+```csharp
+// Define policy modules
+var modules = new List<PolicyModule>
+{
+    new PolicyModule("policy.rego", @"
+        package example
+        import rego.v1
+        
+        default allow := false
+        allow if input.user == ""admin""
+    ")
+};
+
+// Compile the policy
+using var policy = Compiler.CompilePolicyWithEntrypoint("{}", modules, "data.example.allow");
+
+// Evaluate with different inputs
+var result1 = policy.EvalWithInput(@"{""user"": ""admin""}");  // true
+var result2 = policy.EvalWithInput(@"{""user"": ""guest""}");  // false
+```
+
+### Target-Aware Policy (Azure Policy Style)
+
+```csharp
+// Register target definition
+TargetRegistry.RegisterFromJson(@"{
+    ""name"": ""azure.storage"",
+    ""resource_schema_selector"": ""type"",
+    ""resource_types"": {
+        ""Microsoft.Storage/storageAccounts"": {
+            ""schema"": { /* JSON Schema */ }
+        }
+    }
+}");
+
+// Define policy with target
+var modules = new List<PolicyModule>
+{
+    new PolicyModule("policy.rego", @"
+        package policy
+        import rego.v1
+        
+        __target__ := ""azure.storage""
+        
+        default effect := ""deny""
+        effect := ""allow"" if {
+            input.type == ""Microsoft.Storage/storageAccounts""
+            input.properties.supportsHttpsTrafficOnly == true
+        }
+    ")
+};
+
+// Compile for target
+using var policy = Compiler.CompilePolicyForTarget("{}", modules);
+
+// Evaluate Azure resource
+var resource = @"{
+    ""type"": ""Microsoft.Storage/storageAccounts"",
+    ""properties"": {
+        ""supportsHttpsTrafficOnly"": true
+    }
+}";
+
+var result = policy.EvalWithInput(resource);  // "allow"
+```
+
+### Policy Introspection
+
+```csharp
+// Get policy metadata
+var info = policy.GetPolicyInfo();
+
+Console.WriteLine($"Target: {info.TargetName}");
+Console.WriteLine($"Effect Rule: {info.EffectRule}");
+Console.WriteLine($"Modules: {string.Join(", ", info.ModuleIds)}");
+Console.WriteLine($"Resource Types: {string.Join(", ", info.ApplicableResourceTypes)}");
+
+// Access parameters
+if (info.Parameters != null && info.Parameters.Count > 0)
+{
+    foreach (var parameterSet in info.Parameters)
+    {
+        Console.WriteLine($"Module: {parameterSet.SourceFile}");
+        foreach (var param in parameterSet.Parameters)
+        {
+            Console.WriteLine($"Parameter: {param.Name} ({param.Type})");
+            if (param.Default != null)
+                Console.WriteLine($"  Default: {param.Default}");
+        }
+    }
+}
+```
+
+### Concurrent Evaluation
+
+```csharp
+// CompiledPolicy is thread-safe
+var tasks = Enumerable.Range(0, 100).Select(i => 
+    Task.Run(() => policy.EvalWithInput($@"{{""id"": {i}}}"))
+).ToArray();
+
+var results = await Task.WhenAll(tasks);
+```
+
+## Performance Considerations
+
+### Compilation Overhead
+
+- Policy compilation has significant overhead due to parsing and analysis
+- **Best Practice**: Compile once, reuse many times
+- Consider caching compiled policies for repeated use
+
+### Memory Management
+
+- `CompiledPolicy` manages unmanaged resources
+- **Always** dispose of compiled policies using `using` statements or explicit `Dispose()`
+- Disposal is thread-safe and waits for active evaluations
+
+### Thread Safety
+
+- All classes are thread-safe for concurrent reads/evaluations
+- Registry modifications should be done during initialization
+- No external synchronization required
+
+## Error Handling
+
+All methods throw `Exception` on errors with descriptive messages:
+
+```csharp
+try
+{
+    var policy = Compiler.CompilePolicyWithEntrypoint(data, modules, rule);
+    var result = policy.EvalWithInput(input);
+}
+catch (Exception ex)
+{
+    Console.WriteLine($"Error: {ex.Message}");
+}
+```
+
+## Feature Flags
+
+Some functionality requires specific Rust feature flags:
+
+- **azure_policy**: Required for target-aware compilation and policy parameters
+- Without this feature, target-related methods will not be available
+
+## Version Compatibility
+
+- Requires .NET Standard 2.0 or later
+- Compatible with .NET Framework 4.6.1+, .NET Core 2.0+, .NET 5+
+- Uses System.Text.Json for JSON serialization (added as dependency)
+
+## Best Practices
+
+1. **Compile Once, Evaluate Many**: Pre-compile policies for repeated evaluation
+2. **Use Disposable Pattern**: Always dispose of CompiledPolicy instances
+3. **Thread-Safe Design**: Take advantage of built-in thread safety
+4. **Registry Setup**: Configure targets and schemas during application startup
+5. **Error Handling**: Wrap operations in try-catch blocks for robust error handling
+6. **Performance Monitoring**: Monitor evaluation times for performance optimization
+
+## Migration from Engine-Based API
+
+If migrating from an engine-based approach:
+
+```csharp
+// Old approach (if it existed)
+// var engine = new Engine();
+// engine.AddPolicy("policy.rego", policyContent);
+// engine.SetInputJson(inputJson);
+// var result = engine.EvalRule("data.policy.allow");
+
+// New compiled approach
+var modules = new[] { new PolicyModule("policy.rego", policyContent) };
+using var policy = Compiler.CompilePolicyWithEntrypoint("{}", modules, "data.policy.allow");
+var result = policy.EvalWithInput(inputJson);
+```
+
+The compiled approach provides better performance for repeated evaluations and clearer resource management.

bindings/csharp/Benchmarks/Benchmarks.csproj

@@ -0,0 +1,25 @@
+<Project Sdk="Microsoft.NET.Sdk">
+  <PropertyGroup>
+    <OutputType>Exe</OutputType>
+    <TargetFramework>net8.0</TargetFramework>
+    <Nullable>Enable</Nullable>
+  </PropertyGroup>
+  
+  <PropertyGroup>
+    <!-- If the environment variable is set (such as in a Github Action run), append the suffix to the version number -->
+    <RegorusPackageVersionSuffix Condition="'$(VersionSuffix)' != ''">-$(VersionSuffix)</RegorusPackageVersionSuffix>
+  </PropertyGroup>
+
+  <PropertyGroup>
+    <!-- If the environment variable is set (such as in a Github Action run), append the suffix to the version number -->
+    <RegorusPackageVersionSuffix Condition="'$(VersionSuffix)' != ''">-$(VersionSuffix)</RegorusPackageVersionSuffix>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <PackageReference Include="Regorus" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <None Include="../../ffi/target/release/libregorus_ffi.dylib" CopyToOutputDirectory="PreserveNewest" />
+  </ItemGroup>
+</Project>

bindings/csharp/Benchmarks/CompiledPolicyEvaluationBenchmark.cs

@@ -0,0 +1,312 @@
+using System;
+using System.Collections.Generic;
+using System.Diagnostics;
+using System.IO;
+using System.Linq;
+using System.Threading;
+using System.Threading.Tasks;
+using Regorus;
+
+namespace Benchmarks
+{
+    public class CompiledPolicyEvaluationBenchmark
+    {
+        private static readonly string TestDataPath = Path.Combine(
+            Directory.GetCurrentDirectory(),
+            "..", "..", "..",
+            "benches", "evaluation", "test_data"
+        );
+
+        private static readonly (string PolicyFile, string[] InputFiles)[] PolicyInputFiles = new[]
+        {
+            ("rbac_policy.rego", new[] { "rbac_input.json", "rbac_input2.json", "rbac_input3.json" }),
+            ("api_access_policy.rego", new[] { "api_access_input.json", "api_access_input2.json", "api_access_input3.json" }),
+            ("data_sensitivity_policy.rego", new[] { "data_sensitivity_input.json", "data_sensitivity_input2.json", "data_sensitivity_input3.json" }),
+            ("time_based_policy.rego", new[] { "time_based_input.json", "time_based_input2.json", "time_based_input3.json" }),
+            ("data_processing_policy.rego", new[] { "data_processing_input.json", "data_processing_input2.json", "data_processing_input3.json" }),
+            ("azure_vm_policy.rego", new[] { "azure_vm_input.json", "azure_vm_input2.json", "azure_vm_input3.json" }),
+            ("azure_storage_policy.rego", new[] { "azure_storage_input.json", "azure_storage_input2.json", "azure_storage_input3.json" }),
+            ("azure_keyvault_policy.rego", new[] { "azure_keyvault_input.json", "azure_keyvault_input2.json", "azure_keyvault_input3.json" }),
+            ("azure_nsg_policy.rego", new[] { "azure_nsg_input.json", "azure_nsg_input2.json", "azure_nsg_input3.json" })
+        };
+
+        private static readonly string[] PolicyNames = new[]
+        {
+            "rbac_policy",
+            "api_access_policy",
+            "data_sensitivity_policy",
+            "time_based_policy",
+            "data_processing_policy",
+            "azure_vm_policy",
+            "azure_storage_policy",
+            "azure_keyvault_policy",
+            "azure_nsg_policy"
+        };
+
+        private static List<(string Policy, string[] Inputs)> LoadPoliciesWithInputs()
+        {
+            var result = new List<(string Policy, string[] Inputs)>();
+
+            foreach (var (policyFile, inputFiles) in PolicyInputFiles)
+            {
+                var policyPath = Path.Combine(TestDataPath, "policies", policyFile);
+                var policy = File.ReadAllText(policyPath);
+
+                var inputs = inputFiles.Select(inputFile =>
+                {
+                    var inputPath = Path.Combine(TestDataPath, "inputs", inputFile);
+                    return File.ReadAllText(inputPath);
+                }).ToArray();
+
+                result.Add((policy, inputs));
+            }
+
+            return result;
+        }
+
+        private static List<CompiledPolicy> PrepareSharedCompiledPolicies()
+        {
+            var policiesWithInputs = LoadPoliciesWithInputs();
+            var compiledPolicies = new List<CompiledPolicy>();
+
+            foreach (var (policy, _) in policiesWithInputs)
+            {
+                var modules = new[] { new PolicyModule("policy.rego", policy) };
+                var compiled = Compiler.CompilePolicyWithEntrypoint("{}", modules, "data.bench.allow");
+                compiledPolicies.Add(compiled);
+            }
+
+            return compiledPolicies;
+        }
+
+        public static void RunCompiledPolicyEvaluationBenchmark()
+        {
+            var cpuCount = Environment.ProcessorCount;
+            var maxThreads = cpuCount * 2;
+            var threadCounts = new List<int> { 1, 2 };
+
+            // Add even numbers from 4 to maxThreads
+            for (int i = 4; i <= maxThreads; i += 2)
+            {
+                threadCounts.Add(i);
+            }
+
+            Console.WriteLine($"Running compiled policy benchmark with max_threads: {maxThreads}");
+            Console.WriteLine($"Testing with thread counts: {string.Join(", ", threadCounts)}");
+            Console.WriteLine();
+
+            // Benchmark both shared policies and per-iteration compilation
+            var configurations = new[]
+            {
+                (true, "compiled_shared_policies"),
+                (false, "compiled_per_iteration")
+            };
+
+            foreach (var (useSharedPolicies, groupName) in configurations)
+            {
+                Console.WriteLine($"=== {groupName} ===");
+
+                foreach (var threads in threadCounts)
+                {
+                    RunCompiledPolicyBenchmark(threads, useSharedPolicies, groupName);
+                }
+                Console.WriteLine();
+            }
+        }
+
+        public static void RunCompiledPolicyBenchmark(int threads, bool useSharedPolicies, string groupName)
+        {
+            const int warmupSeconds = 3;
+            const int durationSeconds = 3;
+            var policiesWithInputs = LoadPoliciesWithInputs();
+            List<CompiledPolicy>? compiledPolicies = null;
+
+            if (useSharedPolicies)
+            {
+                compiledPolicies = PrepareSharedCompiledPolicies();
+            }
+
+            Console.WriteLine($"Warming up with {threads} threads for {warmupSeconds} seconds...");
+
+            // Warmup phase
+            var (_, _, _, _) = RunBenchmarkPhase(threads, warmupSeconds, policiesWithInputs, compiledPolicies, useSharedPolicies, isWarmup: true);
+
+            Console.WriteLine($"Running benchmark with {threads} threads for {durationSeconds} seconds...");
+
+            // Actual benchmark phase
+            var (totalEvaluations, evaluationTime, policyCounters, allocatedBytes) = RunBenchmarkPhase(threads, durationSeconds, policiesWithInputs, compiledPolicies, useSharedPolicies, isWarmup: false);
+
+            // Calculate throughput based on pure evaluation time (consistent with Rust benchmark)
+            var evalsPerSecond = totalEvaluations / evaluationTime.TotalSeconds;
+            var kelemsPerSecond = evalsPerSecond / 1000.0;
+
+            Console.WriteLine($"{groupName}/eval/{threads} threads");
+            Console.WriteLine($"                        time:   [{evaluationTime.TotalMilliseconds:F2} ms]");
+            Console.WriteLine($"                        thrpt:  [{kelemsPerSecond:F2} Kelem/s]");
+
+            if (totalEvaluations > 0)
+            {
+                var bytesPerEval = allocatedBytes / (double)totalEvaluations;
+                Console.WriteLine($"                        alloc:  [{bytesPerEval:F2} B/op] (total {allocatedBytes} B)");
+            }
+
+            // Clean up compiled policies if we created them
+            if (compiledPolicies != null)
+            {
+                foreach (var policy in compiledPolicies)
+                {
+                    DisposeCompiledPolicy(policy);
+                }
+            }
+
+            // Verify that all policies were evaluated
+            var allEvaluated = policyCounters.Values.All(count => count > 0);
+
+            if (allEvaluated)
+            {
+                Console.WriteLine("✓ All policies were evaluated successfully");
+            }
+            else
+            {
+                Console.WriteLine("ERROR: Some policies were never evaluated successfully!");
+            }
+        }
+
+        private static (int totalEvaluations, TimeSpan evaluationTime, Dictionary<string, int> policyCounters, long allocatedBytes) RunBenchmarkPhase(
+            int threads,
+            int durationSeconds,
+            List<(string Policy, string[] Inputs)> policiesWithInputs,
+            List<CompiledPolicy>? compiledPolicies,
+            bool useSharedPolicies,
+            bool isWarmup)
+        {
+            var barrier = new Barrier(threads);
+            var tasks = new Task[threads];
+            var policyCounters = new Dictionary<string, int>();
+            var evaluationTimes = new Dictionary<int, TimeSpan>();
+            var lockObject = new object();
+            var stopExecution = false;
+            long allocatedBytes = 0;
+
+            // Initialize counters
+            foreach (var policyName in PolicyNames)
+            {
+                policyCounters[policyName] = 0;
+            }
+
+            var stopwatch = Stopwatch.StartNew();
+
+            for (int threadId = 0; threadId < threads; threadId++)
+            {
+                int tid = threadId;
+                tasks[threadId] = Task.Run(() =>
+                {
+                    long allocationStart = 0;
+                    if (!isWarmup)
+                    {
+                        allocationStart = GC.GetAllocatedBytesForCurrentThread();
+                    }
+
+                    barrier.SignalAndWait();
+
+                    int evaluationCount = 0;
+                    var localEvaluationTime = TimeSpan.Zero;
+
+                    while (!stopExecution)
+                    {
+                        // Use different policy for each iteration
+                        int policyIdx = (tid + evaluationCount) % policiesWithInputs.Count;
+                        var (policy, inputs) = policiesWithInputs[policyIdx];
+
+                        // Use different input for the same policy based on iteration
+                        int inputIdx = evaluationCount % inputs.Length;
+                        var input = inputs[inputIdx];
+
+                        try
+                        {
+                            // Measure only the evaluation call
+                            var evalStopwatch = Stopwatch.StartNew();
+
+                            if (useSharedPolicies)
+                            {
+                                var result = compiledPolicies![policyIdx].EvalWithInput(input);
+                            }
+                            else
+                            {
+                                // Compile policy in each iteration.
+                                var modules = new[] { new PolicyModule("policy.rego", policy) };
+                                var compiled = Compiler.CompilePolicyWithEntrypoint("{}", modules, "data.bench.allow");
+                                try
+                                {
+                                    var result = compiled.EvalWithInput(input);
+                                }
+                                finally
+                                {
+                                    DisposeCompiledPolicy(compiled);
+                                }
+                            }
+
+                            evalStopwatch.Stop();
+                            localEvaluationTime += evalStopwatch.Elapsed;
+
+                            // Track successful evaluations (only during actual benchmark, not warmup)
+                            if (!isWarmup)
+                            {
+                                lock (lockObject)
+                                {
+                                    policyCounters[PolicyNames[policyIdx]]++;
+                                }
+                            }
+                        }
+                        catch (Exception)
+                        {
+                            // Ignore evaluation errors for benchmarking purposes
+                        }
+
+                        evaluationCount++;
+                    }
+
+                    // Store the actual evaluation time for this thread
+                    if (!isWarmup)
+                    {
+                        lock (lockObject)
+                        {
+                            if (!evaluationTimes.ContainsKey(tid))
+                                evaluationTimes[tid] = TimeSpan.Zero;
+                            evaluationTimes[tid] = localEvaluationTime;
+                        }
+
+                        var allocationEnd = GC.GetAllocatedBytesForCurrentThread();
+                        System.Threading.Interlocked.Add(ref allocatedBytes, allocationEnd - allocationStart);
+                    }
+                });
+            }
+
+            // Stop execution after the specified duration
+            Task.Delay(TimeSpan.FromSeconds(durationSeconds)).ContinueWith(_ => stopExecution = true);
+
+            Task.WaitAll(tasks);
+            stopwatch.Stop();
+
+            var totalEvaluations = policyCounters.Values.Sum();
+            var totalEvaluationTime = evaluationTimes.Values.Aggregate(TimeSpan.Zero, (sum, time) => sum + time);
+
+            // Use pure evaluation time (consistent with Rust benchmark)
+            var evaluationTime = totalEvaluationTime == TimeSpan.Zero ? stopwatch.Elapsed : totalEvaluationTime;
+
+            return (totalEvaluations, evaluationTime, policyCounters, allocatedBytes);
+        }
+
+        private static void DisposeCompiledPolicy(CompiledPolicy policy)
+        {
+            try
+            {
+                policy.Dispose();
+            }
+            catch (TimeoutException ex)
+            {
+                Console.WriteLine($"Warning: {ex.Message}");
+            }
+        }
+    }
+}

bindings/csharp/Benchmarks/EngineEvaluationBenchmark.cs

@@ -0,0 +1,293 @@
+using System;
+using System.Collections.Generic;
+using System.Diagnostics;
+using System.IO;
+using System.Linq;
+using System.Threading;
+using System.Threading.Tasks;
+using Regorus;
+
+namespace Benchmarks
+{
+    public class EngineEvaluationBenchmark
+    {
+        private static readonly string TestDataPath = Path.Combine(
+            Directory.GetCurrentDirectory(),
+            "..", "..", "..",
+            "benches", "evaluation", "test_data"
+        );
+
+        private static readonly (string PolicyFile, string[] InputFiles)[] PolicyInputFiles = new[]
+        {
+            ("rbac_policy.rego", new[] { "rbac_input.json", "rbac_input2.json", "rbac_input3.json" }),
+            ("api_access_policy.rego", new[] { "api_access_input.json", "api_access_input2.json", "api_access_input3.json" }),
+            ("data_sensitivity_policy.rego", new[] { "data_sensitivity_input.json", "data_sensitivity_input2.json", "data_sensitivity_input3.json" }),
+            ("time_based_policy.rego", new[] { "time_based_input.json", "time_based_input2.json", "time_based_input3.json" }),
+            ("data_processing_policy.rego", new[] { "data_processing_input.json", "data_processing_input2.json", "data_processing_input3.json" }),
+            ("azure_vm_policy.rego", new[] { "azure_vm_input.json", "azure_vm_input2.json", "azure_vm_input3.json" }),
+            ("azure_storage_policy.rego", new[] { "azure_storage_input.json", "azure_storage_input2.json", "azure_storage_input3.json" }),
+            ("azure_keyvault_policy.rego", new[] { "azure_keyvault_input.json", "azure_keyvault_input2.json", "azure_keyvault_input3.json" }),
+            ("azure_nsg_policy.rego", new[] { "azure_nsg_input.json", "azure_nsg_input2.json", "azure_nsg_input3.json" })
+        };
+
+        private static readonly string[] PolicyNames = new[]
+        {
+            "rbac_policy",
+            "api_access_policy",
+            "data_sensitivity_policy",
+            "time_based_policy",
+            "data_processing_policy",
+            "azure_vm_policy",
+            "azure_storage_policy",
+            "azure_keyvault_policy",
+            "azure_nsg_policy"
+        };
+
+        private static List<(string Policy, string[] Inputs)> LoadPoliciesWithInputs()
+        {
+            var result = new List<(string Policy, string[] Inputs)>();
+
+            foreach (var (policyFile, inputFiles) in PolicyInputFiles)
+            {
+                var policyPath = Path.Combine(TestDataPath, "policies", policyFile);
+                var policy = File.ReadAllText(policyPath);
+
+                var inputs = inputFiles.Select(inputFile =>
+                {
+                    var inputPath = Path.Combine(TestDataPath, "inputs", inputFile);
+                    return File.ReadAllText(inputPath);
+                }).ToArray();
+
+                result.Add((policy, inputs));
+            }
+
+            return result;
+        }
+
+        private static List<Engine> PrepareClonedEngines()
+        {
+            var policiesWithInputs = LoadPoliciesWithInputs();
+            var engines = new List<Engine>();
+
+            foreach (var (policy, _) in policiesWithInputs)
+            {
+                var engine = new Engine();
+                engine.AddPolicy("policy.rego", policy);
+
+                // Warm up the engine to ensure it's fully prepared for evaluation
+                // This prevents each cloned engine from repeating preparation work
+                engine.SetInputJson("{}");
+                try
+                {
+                    engine.EvalRule("data.bench.allow");
+                }
+                catch
+                {
+                    // Ignore warmup errors
+                }
+
+                engines.Add(engine);
+            }
+
+            return engines;
+        }
+
+        public static void RunEngineEvaluationBenchmark()
+        {
+            var cpuCount = Environment.ProcessorCount;
+            var maxThreads = cpuCount * 2;
+            var threadCounts = new List<int> { 1, 2 };
+
+            // Add even numbers from 4 to maxThreads
+            for (int i = 4; i <= maxThreads; i += 2)
+            {
+                threadCounts.Add(i);
+            }
+
+            Console.WriteLine($"Running engine benchmark with max_threads: {maxThreads}");
+            Console.WriteLine($"Testing with thread counts: {string.Join(", ", threadCounts)}");
+            Console.WriteLine();
+
+            // Benchmark both cloned engines and fresh engines
+            var configurations = new[]
+            {
+                (true, "cloned_engines"),
+                (false, "fresh_engines")
+            };
+
+            foreach (var (useClonedEngines, groupName) in configurations)
+            {
+                Console.WriteLine($"=== {groupName} ===");
+
+                foreach (var threads in threadCounts)
+                {
+                    RunEngineEvaluationBenchmark(threads, useClonedEngines, groupName);
+                }
+                Console.WriteLine();
+            }
+        }
+
+        public static void RunEngineEvaluationBenchmark(int threads, bool useClonedEngines, string groupName)
+        {
+            const int warmupSeconds = 3;
+            const int durationSeconds = 3;
+            var policiesWithInputs = LoadPoliciesWithInputs();
+
+            Console.WriteLine($"Warming up with {threads} threads for {warmupSeconds} seconds...");
+
+            // Warmup phase
+            var (_, _, _) = RunBenchmarkPhase(threads, warmupSeconds, policiesWithInputs, useClonedEngines, isWarmup: true);
+
+            Console.WriteLine($"Running benchmark with {threads} threads for {durationSeconds} seconds...");
+
+            // Actual benchmark phase
+            var (totalEvaluations, evaluationTime, policyCounters) = RunBenchmarkPhase(threads, durationSeconds, policiesWithInputs, useClonedEngines, isWarmup: false);
+
+            // Calculate throughput based on pure evaluation time (consistent with Rust benchmark)
+            var evalsPerSecond = totalEvaluations / evaluationTime.TotalSeconds;
+            var kelemsPerSecond = evalsPerSecond / 1000.0;
+
+            Console.WriteLine($"{groupName}/eval/{threads} threads");
+            Console.WriteLine($"                        time:   [{evaluationTime.TotalMilliseconds:F2} ms]");
+            Console.WriteLine($"                        thrpt:  [{kelemsPerSecond:F2} Kelem/s]");
+
+            // Verify that all policies were evaluated
+            var allEvaluated = policyCounters.Values.All(count => count > 0);
+
+            if (allEvaluated)
+            {
+                Console.WriteLine("✓ All policies were evaluated successfully");
+            }
+            else
+            {
+                Console.WriteLine("ERROR: Some policies were never evaluated successfully!");
+            }
+        }
+
+        private static (int totalEvaluations, TimeSpan evaluationTime, Dictionary<string, int> policyCounters) RunBenchmarkPhase(
+            int threads,
+            int durationSeconds,
+            List<(string Policy, string[] Inputs)> policiesWithInputs,
+            bool useClonedEngines,
+            bool isWarmup)
+        {
+            var barrier = new Barrier(threads);
+            var tasks = new Task[threads];
+            var policyCounters = new Dictionary<string, int>();
+            var evaluationTimes = new Dictionary<int, TimeSpan>();
+            var lockObject = new object();
+            var stopExecution = false;
+
+            // Initialize counters
+            foreach (var policyName in PolicyNames)
+            {
+                policyCounters[policyName] = 0;
+            }
+
+            // Pre-create engines if using cloned engines
+            List<Engine>? clonedEngines = null;
+            if (useClonedEngines)
+            {
+                clonedEngines = PrepareClonedEngines();
+            }
+
+            var stopwatch = Stopwatch.StartNew();
+
+            for (int threadId = 0; threadId < threads; threadId++)
+            {
+                int tid = threadId;
+                tasks[threadId] = Task.Run(() =>
+                {
+                    barrier.SignalAndWait();
+
+                    int evaluationCount = 0;
+                    var localEvaluationTime = TimeSpan.Zero;
+
+                    while (!stopExecution)
+                    {
+                        // Use different policy for each iteration
+                        int policyIdx = (tid + evaluationCount) % policiesWithInputs.Count;
+                        var (policy, inputs) = policiesWithInputs[policyIdx];
+
+                        // Use different input for the same policy based on iteration
+                        int inputIdx = evaluationCount % inputs.Length;
+                        var input = inputs[inputIdx];
+
+                        try
+                        {
+                            // Measure only the engine operations
+                            var evalStopwatch = Stopwatch.StartNew();
+
+                            Engine engine;
+                            if (useClonedEngines)
+                            {
+                                engine = clonedEngines![policyIdx].Clone();
+                            }
+                            else
+                            {
+                                engine = new Engine();
+                                engine.AddPolicy("policy.rego", policy);
+                            }
+
+                            engine.SetInputJson(input);
+                            var result = engine.EvalRule("data.bench.allow");
+                            engine.Dispose();
+
+                            evalStopwatch.Stop();
+                            localEvaluationTime += evalStopwatch.Elapsed;
+
+                            // Track successful evaluations (only during actual benchmark, not warmup)
+                            if (!isWarmup)
+                            {
+                                lock (lockObject)
+                                {
+                                    policyCounters[PolicyNames[policyIdx]]++;
+                                }
+                            }
+                        }
+                        catch (Exception)
+                        {
+                            // Ignore evaluation errors for benchmarking purposes
+                        }
+
+                        evaluationCount++;
+                    }
+
+                    // Store the actual evaluation time for this thread
+                    if (!isWarmup)
+                    {
+                        lock (lockObject)
+                        {
+                            if (!evaluationTimes.ContainsKey(tid))
+                                evaluationTimes[tid] = TimeSpan.Zero;
+                            evaluationTimes[tid] = localEvaluationTime;
+                        }
+                    }
+                });
+            }
+
+            // Stop execution after the specified duration
+            Task.Delay(TimeSpan.FromSeconds(durationSeconds)).ContinueWith(_ => stopExecution = true);
+
+            Task.WaitAll(tasks);
+            stopwatch.Stop();
+
+            // Clean up cloned engines if we created them
+            if (clonedEngines != null)
+            {
+                foreach (var engine in clonedEngines)
+                {
+                    engine.Dispose();
+                }
+            }
+
+            var totalEvaluations = policyCounters.Values.Sum();
+            var totalEvaluationTime = evaluationTimes.Values.Aggregate(TimeSpan.Zero, (sum, time) => sum + time);
+
+            // Use pure evaluation time (consistent with Rust benchmark)
+            var evaluationTime = totalEvaluationTime == TimeSpan.Zero ? stopwatch.Elapsed : totalEvaluationTime;
+
+            return (totalEvaluations, evaluationTime, policyCounters);
+        }
+    }
+}

bindings/csharp/Benchmarks/Program.cs

@@ -0,0 +1,36 @@
+using System;
+
+namespace Benchmarks
+{
+    class Program
+    {
+        static void Main(string[] args)
+        {
+            Console.WriteLine("=== Regorus C# Benchmarks ===\n");
+
+            try
+            {
+                Console.WriteLine("Running Engine Evaluation Benchmark...");
+                EngineEvaluationBenchmark.RunEngineEvaluationBenchmark();
+            }
+            catch (Exception ex)
+            {
+                Console.WriteLine($"Engine benchmark failed: {ex.Message}");
+            }
+
+            Console.WriteLine("\n" + new string('=', 80) + "\n");
+
+            try
+            {
+                Console.WriteLine("Running Compiled Policy Evaluation Benchmark...");
+                CompiledPolicyEvaluationBenchmark.RunCompiledPolicyEvaluationBenchmark();
+            }
+            catch (Exception ex)
+            {
+                Console.WriteLine($"Compiled policy benchmark failed: {ex.Message}");
+            }
+
+            Console.WriteLine("\n=== Benchmarks Complete ===");
+        }
+    }
+}

bindings/csharp/Benchmarks/compiled_policy_evaluation_benchmark.md

@@ -0,0 +1,140 @@
+# Compiled Policy Evaluation Benchmark Results (C#/.NET)
+
+## Test Environment
+- **Platform**: Apple Silicon (M-Series)
+- **CPU**: 16 cores
+- **Architecture**: ARM64 (aarch64-apple-darwin)
+- **.NET Version**: 8.0
+- **Allocator**: mimalloc (default allocator for Rust FFI)
+- **Benchmark Framework**: Custom time-based benchmarking
+- **Test Data**: 20,000 inputs per evaluation (distributed across threads)
+- **Policy**: Complex authorization policy with nested rules
+- **Warmup Duration**: 3 seconds per configuration
+- **Evaluation Duration**: 3 seconds per configuration
+
+## Benchmark Overview
+
+The C# compiled policy evaluation benchmark tests Regorus compiled policy performance across multiple thread configurations (1-32 threads). It measures throughput (thousands of evaluations per second) for different combinations of compiled policy compilation strategies.
+
+## Configuration Combinations
+
+1. **Compiled Shared Policies**: All threads share pre-compiled policy instances - optimal for performance
+2. **Compiled Per Iteration**: Each thread compiles the policy for each evaluation iteration
+
+*Note: The C# implementation uses a simpler configuration model compared to Rust, which also varies input data handling (cloned vs fresh inputs). The C# benchmarks focus on compilation strategies with consistent input handling.*
+
+## Performance Results
+
+### Compiled Shared Policies (Best Performance)
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                    2905.41 |                  273 |
+|       2 |                    5808.07 |                  240 |
+|       4 |                   11631.23 |                  227 |
+|       6 |                   17431.95 |                  216 |
+|       8 |                   23183.42 |                  126 |
+|      10 |                   28886.11 |                  118 |
+|      12 |                   34659.87 |                  108 |
+|      14 |                   40564.07 |                   84 |
+|      16 |                   46446.38 |                   72 |
+|      18 |                   52047.06 |                   63 |
+|      20 |                   56983.45 |                   58 |
+|      22 |                  404931.47 |                   55 |
+|      24 |                   61673.71 |                   55 |
+|      26 |                   64370.41 |                   51 |
+|      28 |                   56897.04 |                   59 |
+|      30 |                  406850.06 |                   52 |
+|      32 |                   56786.24 |                   58 |
+
+### Compiled Per Iteration
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                    2978.06 |                   49 |
+|       2 |                    5965.09 |                   47 |
+|       4 |                   11928.23 |                   46 |
+|       6 |                   17892.58 |                   45 |
+|       8 |                   23773.82 |                   43 |
+|      10 |                   29705.61 |                   42 |
+|      12 |                   35631.97 |                   40 |
+|      14 |                   41563.35 |                   34 |
+|      16 |                   47452.93 |                   31 |
+|      18 |                   53505.42 |                   27 |
+|      20 |                   59393.86 |                   25 |
+|      22 |                  436115.28 |                   23 |
+|      24 |                   71088.08 |                   21 |
+|      26 |                   76928.70 |                   19 |
+|      28 |                   82759.27 |                   18 |
+|      30 |                  560658.97 |                   17 |
+|      32 |                   93949.39 |                   16 |
+
+## Analysis
+
+The C# compiled policy benchmark demonstrates important performance characteristics with mimalloc as the default allocator:
+
+1. **Compilation Strategy Impact**: Shared compiled policies significantly outperform per-iteration compilation (~5.6x at 1 thread)
+2. **Scaling Patterns with mimalloc**: 
+   - Best throughput achieved at 1 thread for shared policies
+   - Performance generally degrades with increased thread count, but mimalloc provides better allocation efficiency
+3. **Performance Hierarchy**:
+   - Shared compiled policies: Best performance (optimal configuration)
+   - Per-iteration compilation: ~82% reduction from optimal
+4. **Compilation Overhead**: Per-iteration compilation creates substantial overhead, similar to fresh engine creation
+5. **Thread Contention**: Significant performance degradation beyond 8 threads for both configurations, though mimalloc helps mitigate some allocation-related issues
+
+## Comparison with Rust Compiled Policy Evaluation
+
+### Multi-Thread Performance Comparison
+
+| Configuration    | 1 Thread (Kelem/s) | 4 Threads (Kelem/s) | 8 Threads (Kelem/s) |
+|:-----------------|:-------------------|:--------------------|:--------------------|
+|                  | C# / Rust          | C# / Rust           | C# / Rust           |
+| Shared Policies  | 273 / 426          | 227 / 342           | 126 / 185           |
+| Per-iteration    | 49 / 55            | 46 / 50             | 43 / 50             |
+
+### Threading Efficiency Analysis
+
+| Configuration    | Low Contention (1-4t) | Medium Contention (6-12t) | High Contention (16+t) |
+|:-----------------|:----------------------|:--------------------------|:-----------------------|
+|                  | Avg C# / Rust         | Avg C# / Rust             | Avg C# / Rust          |
+| Shared Policies  | 249 / 384             | 150 / 203                 | 58 / 123               |
+| Per-iteration    | 47 / 54               | 40 / 50                   | 22 / 42                |
+
+**Key Observations:**
+- **Single-threaded performance**: C# achieves 64% of Rust performance for shared policies, 89% for per-iteration
+- **Threading scaling**: Both platforms show similar degradation patterns, but Rust maintains better absolute performance
+- **Contention resistance**: Per-iteration compilation shows more consistent relative performance across thread counts
+- **Platform differences**: C# shows more pronounced performance drops at higher thread counts, particularly for shared policies
+
+*Note: Rust benchmarks include additional input data variations (cloned vs fresh inputs) that are not present in the C# implementation.*
+
+## Comparison with C# Engine Evaluation
+
+### Multi-Thread Performance Comparison
+
+| Configuration   | 1 Thread (Kelem/s) | 4 Threads (Kelem/s) | 8 Threads (Kelem/s) |
+|:----------------|:-------------------|:--------------------|:--------------------|
+|                 | CP / EE            | CP / EE             | CP / EE             |
+| Shared Policies | 273 / 279          | 227 / 217           | 126 / 114           |
+| Per-iteration   | 49 / 50            | 46 / 47             | 43 / 45             |
+
+### Threading Efficiency Analysis
+
+| Configuration   | Low Contention (1-4t) | Medium Contention (6-12t) | High Contention (16+t) |
+|:----------------|:----------------------|:--------------------------|:-----------------------|
+|                 | Avg CP / EE           | Avg CP / EE               | Avg CP / EE            |
+| Shared Policies | 249 / 248             | 150 / 128                 | 58 / 54                |
+| Per-iteration   | 47 / 48               | 40 / 39                   | 22 / 27                |
+
+**Key Observations:**
+- **Single-threaded parity**: Both systems perform nearly identically at 1 thread
+- **Threading behavior**: Compiled policies slightly outperform engine evaluation at higher thread counts for shared policies
+- **Contention resistance**: Per-iteration configurations show very similar performance characteristics across all thread counts
+- **Platform consistency**: Both C# implementations show similar scaling patterns and contention behavior
+
+## Performance Insights
+
+1. **C# vs Rust Performance**: C# compiled policies achieve 65% average performance of Rust for shared policies, 87% average for per-iteration across low contention scenarios
+2. **Engine vs Compiled**: In C#, engine and compiled policy evaluation show very similar average performance (compiled policies achieve 100% of engine performance for shared policies, 98% for per-iteration)
+3. **mimalloc Impact**: The use of mimalloc as the default allocator in the underlying Rust FFI provides better memory allocation efficiency and improved threading characteristics
+4. **Threading Scaling**: Both C# configurations demonstrate similar contention patterns, with shared policies showing more pronounced degradation under high thread contention compared to per-iteration compilation
+

bindings/csharp/Benchmarks/engine_evaluation_benchmark.md

@@ -0,0 +1,118 @@
+# Engine Evaluation Benchmark Results (C#/.NET)
+
+## Test Environment
+- **Platform**: Apple Silicon (M-Series)
+- **CPU**: 16 cores
+- **Architecture**: ARM64 (aarch64-apple-darwin)
+- **.NET Version**: 8.0
+- **Allocator**: mimalloc (default allocator for Rust FFI)
+- **Benchmark Framework**: Custom time-based benchmarking
+- **Test Data**: 20,000 inputs per evaluation (distributed across threads)
+- **Policy**: Complex authorization policy with nested rules
+- **Warmup Duration**: 3 seconds per configuration
+- **Evaluation Duration**: 3 seconds per configuration
+
+## Benchmark Overview
+
+The C# engine evaluation benchmark tests Regorus policy evaluation performance across multiple thread configurations (1-32 threads). It measures throughput (thousands of evaluations per second) for different combinations of engine reuse strategies.
+
+## Configuration Combinations
+
+1. **Cloned Engines**: Each thread uses its own cloned engine instance - optimal for performance
+2. **Fresh Engines**: Each thread creates a new engine for each evaluation iteration
+
+*Note: The C# implementation uses a simpler configuration model compared to Rust, which also varies input data handling (cloned vs fresh inputs). The C# benchmarks focus on engine reuse strategies with consistent input handling.*
+
+## Performance Results
+
+### Cloned Engines (Best Performance)
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                    2903.43 |                  279 |
+|       2 |                    5808.35 |                  227 |
+|       4 |                   11645.08 |                  217 |
+|       6 |                   17469.69 |                  207 |
+|       8 |                   23268.07 |                  114 |
+|      10 |                   28996.14 |                  104 |
+|      12 |                   34808.60 |                   98 |
+|      14 |                   40703.21 |                   72 |
+|      16 |                   46488.23 |                   63 |
+|      18 |                   52078.52 |                   56 |
+|      20 |                   57014.31 |                   51 |
+|      22 |                   60482.22 |                   47 |
+|      24 |                   62445.67 |                   46 |
+|      26 |                   65128.74 |                   45 |
+|      28 |                   58001.92 |                   50 |
+|      30 |                   66154.78 |                   42 |
+|      32 |                   64999.03 |                   45 |
+
+### Fresh Engines
+| Threads | Total Evaluation Time (ms) | Throughput (Kelem/s) |
+|--------:|---------------------------:|---------------------:|
+|       1 |                    2982.28 |                   50 |
+|       2 |                    5962.62 |                   48 |
+|       4 |                   11917.94 |                   47 |
+|       6 |                   17874.77 |                   46 |
+|       8 |                   23729.94 |                   45 |
+|      10 |                   29635.17 |                   42 |
+|      12 |                   35574.71 |                   38 |
+|      14 |                   41482.61 |                   34 |
+|      16 |                   47425.16 |                   32 |
+|      18 |                   53248.87 |                   29 |
+|      20 |                   58424.34 |                   27 |
+|      22 |                   61302.24 |                   26 |
+|      24 |                   67430.08 |                   23 |
+|      26 |                   65226.79 |                   24 |
+|      28 |                   73118.48 |                   22 |
+|      30 |                  326472.94 |                   23 |
+|      32 |                   63805.03 |                   24 |
+
+## Analysis
+
+The C# benchmark results demonstrate important performance characteristics with mimalloc as the default allocator:
+
+1. **Engine Reuse Impact**: Cloned engines significantly outperform fresh engines (~5.6x at 1 thread)
+2. **Scaling Patterns with mimalloc**: 
+   - Best throughput achieved at 1 thread for both configurations
+   - Performance degrades with increased thread count due to contention, but mimalloc provides better allocation efficiency
+   - Cloned engines show better relative scaling characteristics
+3. **Performance Hierarchy**: 
+   - Cloned engines: Best performance (optimal configuration)
+   - Fresh engines: ~82% reduction from optimal
+4. **Thread Contention**: Significant performance drop beyond 8 threads, especially for fresh engines, though mimalloc helps mitigate some allocation-related issues
+5. **C# vs Rust Performance**: C# shows ~66% of Rust performance for equivalent cloned engine configuration
+
+## Comparison with Rust Engine Evaluation
+
+### Multi-Thread Performance Comparison
+
+| Configuration  | 1 Thread (Kelem/s) | 4 Threads (Kelem/s) | 8 Threads (Kelem/s) |
+|:---------------|:-------------------|:--------------------|:--------------------|
+|                | C# / Rust          | C# / Rust           | C# / Rust           |
+| Cloned Engines | 279 / 423          | 217 / 406           | 114 / 341           |
+| Fresh Engines  | 50 / 56            | 47 / 54             | 45 / 53             |
+
+### Threading Efficiency Analysis
+
+| Configuration  | Low Contention (1-4t) | Medium Contention (6-12t) | High Contention (16+t) |
+|:---------------|:----------------------|:--------------------------|:-----------------------|
+|                | Avg C# / Rust         | Avg C# / Rust             | Avg C# / Rust          |
+| Cloned Engines | 253 / 414             | 128 / 329                 | 54 / 250               |
+| Fresh Engines  | 48 / 55               | 39 / 52                   | 27 / 42                |
+
+**Key Observations:**
+- **Single-threaded performance**: C# achieves 66% of Rust performance for cloned engines, 89% for fresh engines
+- **Threading scaling**: Both platforms show similar degradation patterns, but Rust maintains better absolute performance
+- **Contention resistance**: Fresh engines show more consistent relative performance across thread counts
+- **Platform differences**: C# shows more pronounced performance drops at higher thread counts, particularly for cloned engines
+
+*Note: Rust benchmarks include additional input data variations (cloned vs fresh inputs) that are not present in the C# implementation.*
+
+## Performance Insights
+
+1. **Engine Creation Overhead**: Fresh engine creation has significant performance impact in C# (~5.6x slower than cloned engines)
+2. **Thread Scaling**: C# shows moderate thread contention with better characteristics when using mimalloc
+3. **Memory Management**: .NET garbage collection patterns combined with mimalloc allocation efficiency
+4. **Interop Performance**: C# bindings achieve 66% of Rust performance for cloned engines, demonstrating effective FFI implementation
+5. **mimalloc Benefits**: The use of mimalloc as the default allocator in the underlying Rust FFI provides improved memory allocation efficiency and better threading characteristics
+

bindings/csharp/Directory.Packages.props

@@ -0,0 +1,15 @@
+<Project>
+  <PropertyGroup>
+    <ManagePackageVersionsCentrally>true</ManagePackageVersionsCentrally>
+    <RegorusPackageVersion>0.9.1</RegorusPackageVersion>
+    <RegorusPackageVersionSuffix Condition="'$(VersionSuffix)' != ''">-$(VersionSuffix)</RegorusPackageVersionSuffix>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <!-- Centralize Regorus package version with optional CI suffix -->
+    <PackageVersion Include="Regorus" Version="$(RegorusPackageVersion)$(RegorusPackageVersionSuffix)" />
+    <PackageVersion Include="MSTest" Version="3.8.2" />
+    <PackageVersion Include="System.Text.Json" Version="8.0.5" />
+    <PackageVersion Include="YamlDotNet" Version="13.7.0" />
+  </ItemGroup>
+</Project>

bindings/csharp/README.md

@@ -0,0 +1,152 @@
+# Regorus CSharp
+
+**Regorus** is
+
+  - *Rego*-*Rus(t)*  - A fast, light-weight [Rego](https://www.openpolicyagent.org/docs/latest/policy-language/)
+   interpreter written in Rust.
+  - *Rigorous* - A rigorous enforcer of well-defined Rego semantics.
+
+See main [Regorus page](https://github.com/microsoft/regorus) for more details about the project.
+
+# Building
+
+## Github Actions
+
+The simplest way to build a Nuget for Regorus' C# bindings is to use Github Actions. The action to do so is named `bindings/csharp` and is defined in `.github/workflows/test-csharp.yml`.
+
+There are two ways to trigger a Nuget build.
+1. Runs are triggered automatically whenever a push or pull request is made to the `main` branch.
+2. A run can be triggered manually by navigating to the action in the Github UI and clicking `Run workflow`. This option allows you to generate a Nuget for any branch, which is useful when testing the integration of in-progress changes to Regorus with other projects. Nuget files that are generated via this flow will have a `manualtrigger` suffix appended to their version number, making it easy to distinguish them from Nugets generated using the `main` branch.
+![Image displaying the run workflow button](docs/images/readme/manuallytriggeringrun.png)
+
+Once the workflow run completes, the generated Nuget can be downloaded by following these steps:
+1. Open the run.
+2. Click on `Build Regorus nuget` on the left.
+3. Expand the `Upload Regorus nuget` step.
+4. Click the `Artifact download URL` link at the bottom.
+5. Save and extract the downloaded zip file to find the `.nupkg` file.
+![Image displaying the download URL link](docs/images/readme/downloadnuget.png)
+
+## Local
+
+The `cargo xtask` runner provides helpers for local builds:
+
+1. `cargo xtask ffi` builds the `bindings/ffi` crate for the host platform in debug mode. Add `--target <triple>` (repeatable) to cross-compile, or `--release` to produce optimised artefacts. Results land under `bindings/ffi/target/<triple>/<profile>`.
+2. `cargo xtask nuget` reuses those artefacts to pack the C# library. It defaults to debug builds for the host but accepts `--target`, `--release`, `--artifacts-dir <path>` to reuse existing binaries, and `--enforce-artifacts` to require every officially supported platform.
+3. `cargo xtask test-csharp` ensures a NuGet is available (rebuilding when required or when `--force-nuget` is passed) and then runs `Regorus.Tests`, `TestApp`, and `TargetExampleApp` against it. The command accepts the same build flags as `cargo xtask nuget`.
+
+## Memory Usage Safeguards
+
+The C# bindings expose allocator-backed memory tracking utilities via the static `Regorus.MemoryLimits` helper. Typical usage:
+
+```csharp
+// Restrict total allocations to 128 MiB for the process
+Regorus.MemoryLimits.SetGlobalMemoryLimit(128 * 1024 * 1024);
+
+// Optional: tune how frequently each thread flushes its allocation counters
+Regorus.MemoryLimits.SetThreadFlushThresholdOverride(256 * 1024);
+
+// Engine operations throw InvalidOperationException with the allocator message if the budget is exceeded
+using var engine = new Regorus.Engine();
+var veryLargeJson = new string('x', 128 * 1024);
+try
+{
+    engine.SetInputJson(veryLargeJson);
+}
+catch (InvalidOperationException ex)
+{
+    Console.WriteLine($"Allocator reported: {ex.Message}");
+}
+
+// Restore defaults once done
+Regorus.MemoryLimits.SetGlobalMemoryLimit(null);
+Regorus.MemoryLimits.SetThreadFlushThresholdOverride(null);
+```
+
+See bindings/csharp/Regorus.Tests/RegorusTests.cs for scenario coverage and bindings/csharp/TargetExampleApp/Program.cs for end-to-end usage.
+
+## RVM Usage Example
+
+The RVM API lets you compile a program from modules/entrypoints and execute it in a VM:
+
+```csharp
+using Regorus;
+
+const string Policy = """
+package demo
+default allow = false
+allow if {
+  input.user == "alice"
+  some role in data.roles[input.user]
+  role == "admin"
+}
+""";
+
+const string Data = """
+{ "roles": { "alice": ["admin"] } }
+""";
+
+const string Input = """
+{ "user": "alice" }
+""";
+
+var modules = new[] { new PolicyModule("demo.rego", Policy) };
+var entryPoints = new[] { "data.demo.allow" };
+
+using var program = Program.CompileFromModules(Data, modules, entryPoints);
+var listing = program.GenerateListing();
+
+using var vm = new Rvm();
+vm.LoadProgram(program);
+vm.SetDataJson(Data);
+vm.SetInputJson(Input);
+
+var result = vm.Execute();
+Console.WriteLine($"allow: {result}");
+```
+
+## Azure RBAC Condition Evaluation
+
+Evaluate Azure RBAC condition expressions directly with a JSON evaluation context:
+
+```csharp
+using Regorus;
+
+const string Condition = "@Resource[owner] StringEquals 'alice'";
+const string ContextJson = """
+{
+  "principal": {
+    "id": "user-1",
+    "principal_type": "User",
+    "custom_security_attributes": {}
+  },
+  "resource": {
+    "id": "/subscriptions/s1",
+    "resource_type": "Microsoft.Storage/storageAccounts",
+    "scope": "/subscriptions/s1",
+    "attributes": {
+      "owner": "alice",
+      "confidential": true
+    }
+  },
+  "request": {
+    "action": "Microsoft.Storage/storageAccounts/read",
+    "data_action": null,
+    "attributes": {
+      "clientIP": "10.0.0.1"
+    }
+  },
+  "environment": {
+    "is_private_link": null,
+    "private_endpoint": null,
+    "subnet": null,
+    "utc_now": "2023-05-01T12:00:00Z"
+  },
+  "action": "Microsoft.Storage/storageAccounts/read",
+  "suboperation": null
+}
+""";
+
+var allowed = RbacEngine.EvaluateCondition(Condition, ContextJson);
+Console.WriteLine($"RBAC condition allowed: {allowed}");
+```