OrygnsCode

OPA MCP Server

Community OrygnsCode
Updated

Model Context Protocol server for Open Policy Agent (OPA). Author, validate, debug, and explain Rego policies through any MCP-compatible client.

OPA MCP Server

CICodeQLnpm versionDocker pullsLicense: MITNode.js

A Model Context Protocol (MCP) serverthat turns any MCP-compatible client (Claude Desktop, Claude Code, Cursor,VS Code, Windsurf, Zed, and others) into a first-classOpen Policy Agent and Rego authoringenvironment.

┌────────────────────┐  MCP / stdio  ┌─────────────────┐  spawn / HTTP  ┌──────────────────┐
│  Claude · Cursor · │ ────────────▶ │  @orygn/opa-mcp │ ─────────────▶ │  opa · regal ·   │
│   VS Code · ...    │ ◀──────────── │                 │ ◀───────────── │  OPA REST API    │
└────────────────────┘   32 tools    └─────────────────┘                └──────────────────┘

Status: v0.1.0. First stable release. Tool surface, error codes, andenvironment variables follow SemVer from thisversion forward.

Table of contents

  • What you can do with it
  • Why this MCP
  • Install
  • Configuration
  • Tool reference
  • Prompts
  • Resources
  • Cookbook
  • Architecture
  • Security
  • Troubleshooting
  • Development
  • Versioning & support
  • License

What you can do with it

Once an MCP client is connected, an agent can:

  • Author Rego. Generate, format, and refactor policies. The serverruns the real opa fmt and opa parse so output is byte-identical towhat you'd get on the command line, and regal (optional) surfacesidiomatic suggestions.
  • Evaluate against data. Run a query against a policy and an inputdocument. Optional --explain, --profile, and --coverage flagssurface execution traces, hot rules, and per-line coverage.
  • Debug a deny. rego_explain_decision walks the agent through everyrule that fired (and every one that didn't), so it can answer "why wasthis rejected" without you reading the trace by hand.
  • Manage policies on a running OPA. List, get, put, delete policies onan OPA server through its REST API. Works against a localopa run --server or a production deployment with bearer-token auth.
  • Build & sign bundles. Package a directory of policies into adeployable bundle, optionally signing it. Output is a regular .tar.gzthe agent can hand to your delivery system.
  • Lint. rego_lint runs Regal across a directory or a single fileand returns categorized findings (style, bugs, performance, idioms).

A walk-through of a typical session lives in Cookbook.

Why this MCP

OPA already has a perfectly good CLI and REST API. So why an MCP wrapper?

  • Schema-shaped tool surface. An agent calling rego_eval gets avalidated input schema, a structured output envelope, and stable errorcodes, instead of parsing free-form CLI text and inventing its ownfailure taxonomy. That alone makes Rego usable to an agent the way alanguage server makes a language usable to an IDE.
  • Higher-level helpers. rego_explain_decision,rego_generate_test_skeleton, rego_describe_policy, andrego_suggest_fix compose the lower-level primitives into the tasksagents are actually asked to do. They don't exist in the OPA CLI.
  • Curated knowledge. The bundled MCP resources expose the OPAbuilt-in function catalog, the official Rego style guide (formatted forLLMs), and a curated pattern library covering RBAC, ABAC, Kubernetesadmission, IaC gates, API authz, and rate limiting, so the agent hasauthoritative context without needing to scrape it.
  • Safety boundaries the agent can rely on. Path allow-list,subprocess timeouts, response-size caps, and an explicitHTTP_SEND_BLOCKED error for the dangerous OPA built-ins. Defaults areconservative; running the server doesn't quietly grant the agent morereach than the operator intended.

If you've ever watched an agent fight opa eval's argument order, you'llrecognize the gap this fills.

Install

The server runs locally over stdio. Pick the install path that matchesyour client.

Claude Desktop / Claude Code

The fastest path is the Smithery one-liner:

npx -y @smithery/cli install @orygn/opa-mcp --client claude

Or download opa-mcp.mcpb from thelatest releaseand double-click it.

If you prefer to edit claude_desktop_config.json by hand, the snippetlives in examples/claude-desktop.json:

{
  "mcpServers": {
    "opa": {
      "command": "npx",
      "args": ["-y", "@orygn/opa-mcp"],
      "env": {
        "OPA_BINARY": "/usr/local/bin/opa",
        "REGAL_BINARY": "/usr/local/bin/regal",
        "OPA_URL": "http://localhost:8181",
        "OPA_MCP_ALLOWED_PATHS": "/path/to/your/policies"
      }
    }
  }
}

Replace the /usr/local/bin/... paths with your real ones. See thefirst-time install gotchabelow. Windows users substitute C:\\path\\to\\opa.exe.

Cursor

Drop examples/cursor.json into either.cursor/mcp.json (project-scoped) or ~/.cursor/mcp.json (user-scoped).

VS Code (GitHub Copilot Chat)

Drop examples/vscode.json into.vscode/mcp.json, or paste the servers block into your usersettings.json under mcp.servers.

Windsurf, Zed, and others

See examples/ for a full set of drop-in configs.

Manual install (any MCP client)

npm install -g @orygn/opa-mcp
opa-mcp --version

then point your client at the opa-mcp binary.

Docker

docker pull orygn/opa-mcp:latest
docker run --rm -i \
  -v /path/to/your/policies:/policies:ro \
  -e OPA_MCP_ALLOWED_PATHS=/policies \
  orygn/opa-mcp

The image is multi-arch (linux/amd64, linux/arm64), bundles pinnedversions of opa and regal, and runs as a non-root user. No hostinstall of OPA or Regal is required.

⚠ First-time install gotcha (read this if you used npx or the global install)

If your client's PATH doesn't include the directory where opa lives(this happens with Claude Desktop on Windows and macOS by default), theserver boots fine but every tool call returns OPA_BINARY_NOT_FOUND.

Fix: add OPA_BINARY and REGAL_BINARY env entries to your clientconfig with the absolute path to each binary. The example configs underexamples/ ship with placeholder paths you replace.Find the real paths with:

which opa && which regal                                    # macOS / Linux
Get-Command opa, regal | Select-Object Source              # Windows

This does not affect the Docker or MCPB install paths; thoseship opa and regal inside the bundle and bypass PATH entirely.See Troubleshooting for full detail.

Configuration

The server reads its configuration from environment variables. Everyvariable is optional; defaults are sensible for a local OPA onhttp://localhost:8181.

Variable Default Purpose
OPA_URL http://localhost:8181 Base URL of an OPA REST endpoint, used by opa_* tools.
OPA_TOKEN (unset) Bearer token for OPA, if your instance requires auth. Treated as a secret. Never echoed in logs or tool responses.
OPA_BINARY opa (on PATH) Path to the opa CLI, used by rego_* tools.
REGAL_BINARY regal (on PATH) Path to the regal linter. Only required by rego_lint.
OPA_MCP_ALLOWED_PATHS (unset) Comma- or semicolon-separated list of directories the server is allowed to read policies from. When unset, file-based tools refuse to read from disk.
OPA_MCP_LOG_FILE <tmpdir>/orygn-opa-mcp.log Path the server appends logs to. The server never writes to stdout; that channel is reserved for the MCP protocol.
OPA_MCP_LOG_LEVEL info One of debug, info, warn, error.
OPA_MCP_MAX_RESPONSE_BYTES 100000 Hard cap on a single tool response. Larger payloads are truncated with a __truncated: true marker.
OPA_MCP_TIMEOUT_MS 30000 Hard timeout for any spawned subprocess (opa, regal). After this, the child gets SIGTERM and then SIGKILL.
OPA_MCP_HTTP_TIMEOUT_MS 15000 Timeout for HTTP requests to the OPA REST API.

Paths in OPA_MCP_ALLOWED_PATHS and the *_BINARY variables must beabsolute. Relative paths and missing binaries are rejected with structurederrors.

Tool reference

Every tool returns a JSON envelope:

{ "ok": true, "data": { ... }, "warnings": [ ... ] }
{ "ok": false, "error": { "code": "INVALID_REGO", "message": "...", "hint": "...", "details": { ... } } }

Stable error codes: INVALID_INPUT, INVALID_REGO, INVALID_BUNDLE,EVAL_ERROR, OPA_BINARY_NOT_FOUND, REGAL_NOT_FOUND,REGAL_VERSION_TOO_OLD, OPA_UNREACHABLE, OPA_AUTH_FAILED,POLICY_NOT_FOUND, PATH_NOT_ALLOWED, PATH_NOT_FOUND,DEPENDENCY_CONFLICT, NO_TESTS_FOUND, HTTP_SEND_BLOCKED, TIMEOUT,UNKNOWN_ERROR.

Category A: Authoring & static analysis

Operate on Rego source code without needing a running OPA server. Wrapopa fmt, opa parse, opa check, opa inspect, opa capabilities,opa deps, and regal.

Tool What it does
rego_format Format Rego source. Wraps opa fmt. Idempotent.
rego_check Type-check and validate Rego. Wraps opa check.
rego_lint Run Regal across a file or directory. Returns findings grouped by category. Requires regal on PATH or REGAL_BINARY set.
rego_parse_ast Parse Rego to AST JSON. Wraps opa parse.
rego_inspect Inspect a bundle or directory: packages, rules, annotations. Wraps opa inspect.
rego_capabilities Return the capabilities (built-ins, future keywords) understood by the bundled OPA.
rego_deps Static dependency analysis: rule-level data references and cross-package calls.
Featured: rego_format
// Input
{
  "source": "package x\nallow{input.user==\"admin\"}"
}

// Output (ok)
{
  "ok": true,
  "data": {
    "formatted": "package x\n\nallow if input.user == \"admin\"\n",
    "changed": true
  }
}
Featured: rego_check
// Input
{
  "source": "package x\nallow if y",
  "strict": true
}

// Output (error path; the JSON diagnostics arrive on stderr from opa)
{
  "ok": true,
  "data": {
    "valid": false,
    "errors": [
      {
        "code": "rego_unsafe_var_error",
        "message": "var y is unsafe",
        "location": { "row": 2, "col": 11 }
      }
    ]
  }
}

Category B: Evaluation & testing

Run a query against a policy and input. Wrap opa eval, opa test, andopa bench.

Tool What it does
rego_eval Evaluate a query against a policy and input. The bread-and-butter tool.
rego_eval_with_explain Evaluate with --explain=full and return a structured trace.
rego_eval_with_profile Evaluate with --profile and return per-rule timing and evaluation counts.
rego_eval_with_coverage Evaluate with --coverage and return per-line coverage.
rego_test Run opa test over a directory. Returns pass/fail per test, with optional coverage.
rego_bench Run opa bench and return statistical timing data.
rego_compile_query Partially evaluate a query against a policy.
Featured: rego_eval
// Input
{
  "query": "data.rbac.allow",
  "source": "package rbac\nimport rego.v1\nallow if input.role == \"admin\"",
  "input": { "role": "admin" }
}

// Output
{
  "ok": true,
  "data": {
    "result": [{ "expressions": [{ "value": true, "text": "data.rbac.allow", "location": { "row": 1, "col": 1 } }] }]
  }
}

Category C: Bundle operations

Package and sign deployable bundles. Wrap opa build and opa sign.

Tool What it does
opa_bundle_build Build a .tar.gz bundle from a policy directory. Supports optimize and revision.
opa_bundle_sign Sign a bundle with a private key. Returns .signatures.json content.

Category D: OPA server management

Talk to a running OPA server over its REST API. Require OPA_URL topoint at a reachable server.

Tool What it does
opa_list_policies List policies registered on the server.
opa_get_policy Get a single policy by ID.
opa_put_policy Upload or replace a policy.
opa_delete_policy Delete a policy by ID.
opa_get_data Read a path from the data hierarchy.
opa_put_data Write to a path in the data hierarchy.
opa_patch_data Apply a JSON Patch to the data hierarchy.
opa_query_decision POST to a /v1/data/... decision endpoint with input.
opa_compile_query Partially evaluate a query against the running server.
opa_health Liveness / readiness check.
opa_status Bundle / decision-log status.
opa_config Server configuration (without secrets).

Category E: Higher-level helpers

The differentiation surface. These compose lower-level primitives intothe tasks agents are actually asked to do.

Tool What it does
rego_explain_decision Walk through every rule that fired (and didn't) for a given query. Wraps rego_eval_with_explain and produces a step-by-step natural-language trace.
rego_generate_test_skeleton Given a policy, generate a _test.rego skeleton covering each rule.
rego_describe_policy Summarize what a policy does, its inputs, decisions, and assumptions.
rego_suggest_fix For a failed rego_check or rego_lint, propose minimal patches.

Prompts

Three MCP promptsship with the server. Clients surface them as slash commands or workflowtemplates.

Prompt Purpose
policy_authoring_assistant Walks the agent through writing a new policy: ask about the decision surface, draft, review, format, lint, test.
policy_review_checklist Review checklist for an existing policy: completeness, edge cases, performance, security pitfalls.
decision_debugging_workflow Diagnostic flow when a decision is unexpected: gather input, run with explain, isolate the rule, propose a fix.

Resources

Three MCP resourcesexpose curated reference data the agent can read at any time.

Resource URI What's there
opa://builtins Categorized OPA built-in function reference, derived at read time from opa capabilities --current. Security-sensitive functions (http.send, crypto.x509.*, opa.runtime) are flagged.
opa://style-guide Condensed Rego style guide, formatted for LLM consumption.
opa://patterns Curated common-pattern library: RBAC, ABAC, Kubernetes admission, IaC gates, API authz, rate limiting. Each pattern includes when-to-use, full Rego, a test, and common pitfalls.

Cookbook

A few session shapes that the tool set was designed for.

"Help me write a policy"

You: I need an authz policy: editors can read/write, viewers can only read,
     admins can do anything.

Agent: I'll draft it. (calls rego_format on a draft, then rego_check, then
       rego_lint)

Agent: Here's the policy. I've also generated a test file with cases for
       each role. (calls rego_generate_test_skeleton, then rego_test)

Agent: All 9 tests pass. Want me to save it to <path>?

"Why was this denied?"

You: This API call is being denied and I don't know why.
     [pastes input.json]

Agent: (calls rego_explain_decision against your local policy with that input)

Agent: The deny comes from rule `forbid_anonymous_writes` at line 17.
       Specifically, `input.user` is null and the request method is "POST".
       The rule fires, which causes the default deny. To allow this, you'd
       need either an authenticated user or a policy exception for this
       endpoint.

"Push this policy to staging OPA"

You: Push policies/rbac.rego to the staging OPA server, but first lint and
     test it.

Agent: (rego_lint → 2 style warnings, no errors)
       (rego_test on policies/ → all pass)
       (opa_put_policy with id="rbac" against $OPA_URL)
       (opa_get_policy to verify)

Agent: Done. Policy `rbac` is live on staging at $OPA_URL.

Architecture

┌──────────────────────────────────── @orygn/opa-mcp ───────────────────────────────────┐
│                                                                                       │
│   src/server.ts ──── McpServer (stdio) ─── tool / prompt / resource registries        │
│                          │                                                            │
│                          ├── tools/authoring/         ─┐                              │
│                          ├── tools/evaluation/        ─┤                              │
│                          ├── tools/bundles/           ─┼─── lib/opa-cli.ts ──┐        │
│                          ├── tools/server-management/ ─┤                     │        │
│                          ├── tools/helpers/           ─┘                     │        │
│                          │                                                   ▼        │
│                          │                              lib/subprocess.ts ──┴── opa   │
│                          │                              lib/regal-cli.ts   ───── regal│
│                          │                              lib/opa-client.ts  ───── HTTP │
│                          │                                                            │
│                          └── lib/output.ts (envelope + truncation)                    │
│                              lib/security.ts (path allow-list)                        │
│                              lib/errors.ts (structured failures)                      │
│                              lib/logger.ts (file-only, never stdout)                  │
└───────────────────────────────────────────────────────────────────────────────────────┘

Three things worth knowing if you're going to operate this:

  1. stdout is the protocol channel. The server logs to a file vialib/logger.ts and never writes to stdout. If you see stray stdoutbytes, the client disconnects; the MCP transport layer is strict.
  2. No tool throws. Every tool catches its own exceptions and returnsa structured { ok: false, error: ... } envelope. The agent sees astable error vocabulary, not a stack trace.
  3. Subprocesses are tightly bounded. lib/subprocess.ts runs opaand regal with shell: false, a hard timeout, and SIGTERM-then-SIGKILL escalation. There is no path through the server where anagent can construct a shell command.

Security

This server is designed to run locally, started by an MCP client onthe user's own machine, communicating over stdio. It is not designed tobe exposed on the network.

  • File-based tools refuse to read anything outside OPA_MCP_ALLOWED_PATHS.When that variable is unset, file tools return PATH_NOT_ALLOWED.
  • Subprocesses run with shell: false and a hard timeout.
  • OPA_TOKEN is never echoed in tool responses or log entries.
  • Releases are published withnpm provenance;the Docker image is built reproducibly from the committed Dockerfile.

To report a vulnerability, follow SECURITY.md. Pleasedo not open a public issue for security problems.

Troubleshooting

Common issues, fast fixes.

OPA_BINARY_NOT_FOUND even though opa is installed. (most commonfirst-day issue, read this first)

MCP clients (notably Claude Desktop on Windows and macOS) launch theserver with a deliberately reduced PATH that omits user-local bindirectories, even ones that work fine in your interactive shell. Thebinary is on your machine; the spawned MCP server just can't see it.

Find the absolute path to opa:

# macOS / Linux
which opa
# → /usr/local/bin/opa  (or /opt/homebrew/bin/opa, or ~/.local/bin/opa)
# Windows
Get-Command opa | Select-Object -ExpandProperty Source
# → C:\Users\you\bin\opa.exe  (or wherever)

Then set OPA_BINARY to that absolute path in your client's MCP envblock. Same for REGAL_BINARY if you use the rego_lint tool. Theexamples/ configs already include both env vars; justedit the placeholder paths.

This issue does not affect the Docker or MCPB install paths. Thosebundle opa and regal and bypass PATH entirely.

The server starts, then the client says "disconnected."

The most likely cause is something in the process writing to stdoutbesides MCP frames. If you've added a custom tool, check that no libraryit calls prints to stdout. The fixed-position safety net islib/logger.ts. Use it, not console.log.

PATH_NOT_ALLOWED on a file under my project.

OPA_MCP_ALLOWED_PATHS is empty by default. Set it to the absolutepath(s) you want the server to read from, comma-separated.

OPA_UNREACHABLE when calling opa_* tools.

OPA_URL (default http://localhost:8181) must point at a running OPAserver (opa run --server ...). Check with curl $OPA_URL/health.

Regal "version too old."

We track the current Regal release. If REGAL_VERSION_TOO_OLD fires,upgrade Regal: brew upgrade regal or download from theRegal releases page.

directory-package-mismatch violation when linting inline source.

When you pass source rather than paths to rego_lint, Regal sees arandomized temp-file path that can't possibly match your declared packagepath. The diagnostic is an artifact of inline linting, not a real issue.Disable the rule for inline workflows or lint via paths against thereal on-disk file when you want canonical signal.

Where are the logs?

Default location is <OS-tmpdir>/orygn-opa-mcp.log. That's typically/tmp/orygn-opa-mcp.log on Linux/macOS or %TEMP%\orygn-opa-mcp.logon Windows. Set OPA_MCP_LOG_FILE to override, andOPA_MCP_LOG_LEVEL=debug to widen the firehose.

Development

git clone https://github.com/OrygnsCode/opa-mcp-server.git
cd opa-mcp-server
npm install
npm run dev

Common commands:

npm run lint              # ESLint
npm run typecheck         # tsc --noEmit
npm test                  # unit tests (Vitest)
npm run test:coverage     # unit + coverage report
npm run test:integration  # against real opa + regal binaries
npm run build             # compile to dist/

CI runs lint, typecheck, build, and unit tests on every push and PRacross Ubuntu, macOS, and Windows on Node 20 and 22. Integration testsrun on Linux against pinned opa and regal releases.

For the full contributor workflow (adding tools, naming conventions,logging discipline, release process), see CONTRIBUTING.md.

Versioning & support

This project follows Semantic Versioning. The publicsurface for SemVer purposes is the set of registered tools, prompts, andresources, their input/output schemas, the recognized environmentvariables, and the CLI entry point.

Breaking changes will be:

  • announced in CHANGELOG.md under a new major version,
  • preceded by at least one minor release with a deprecation warning,
  • accompanied by a migration note in the release announcement.

Pinned versions of the upstream toolchain (opa and regal) are treatedas part of the build, not as a dependency the operator manages. TheDockerfile, MCPB bundle, and CI all use the same pin; bumps go throughDependabot or a manual PR.

License

MIT © Orygn LLC

@orygn/opa-mcp is an independent project. It is not affiliated with,endorsed by, or sponsored by the Open Policy Agent project, the CloudNative Computing Foundation, Styra, or Anthropic. "Open Policy Agent"and "Rego" are trademarks of their respective owners. "Model ContextProtocol" is a trademark of Anthropic, PBC.

MCP Server · Populars

MCP Server · New

    cognicore-dev

    CogniCore

    Persistent memory, reflection & safety for AI agents. No vector DB, no embedding server — BM25 + multi-hop retrieval runs in-process. MCP server with 10 tools. Apache 2.0.

    Community cognicore-dev
    avivgerz-hue

    🛡️ vibe-shield

    60-rule static security scanner, MCP server & skill poisoning auditor for AI-generated codebases (Claude Code, Codex, Cursor).

    Community avivgerz-hue
    frankShan001

    CutVoke

    Open-source, local-first video editor your AI agent can drive over MCP (Claude Code, Codex, Cursor). Windows · macOS · Linux. The AI makes the cut; you finish it on the same timeline.

    Community frankShan001
    HiveForensics-AI

    @knolo/mcp

    The open framework AI agents use to share knowledge with one another. Pack knowledge into a portable, verifiable Knowledge Image, pin it by digest, mount it locally over MCP, and prove what was used. Local-first, deterministic, no vector DB required.

    Community HiveForensics-AI
    shannonbay

    Quarterdeck

    Persistent goals for Claude Code, like "keep my inbox under control". Fresh agent runs on a schedule, webhook or failing health check, with instructions, memory and your decisions kept outside the agent. Actions wait for your approval until you trust them, enforced by a hook. MCP server + web app.

    Community shannonbay