<p align="center">
<img src="https://raw.githubusercontent.com/Goldziher/ai-rulez/main/docs/assets/ai-rulez-banner.png" alt="AI-Rulez" width="820" />
</p>
<p align="center">
<strong>The standards-compliant lifecycle tool for agentic governance</strong>
</p>
<p align="center">
<a href="https://www.npmjs.com/package/ai-rulez"><img src="https://img.shields.io/npm/v/ai-rulez?color=blue" alt="npm version"></a>
<a href="https://pypi.org/project/ai-rulez/"><img src="https://img.shields.io/pypi/v/ai-rulez?color=blue" alt="PyPI version"></a>
<a href="https://github.com/Goldziher/homebrew-tap"><img src="https://img.shields.io/github/v/release/Goldziher/ai-rulez?label=homebrew&color=blue" alt="Homebrew version"></a>
<a href="https://pkg.go.dev/github.com/Goldziher/ai-rulez/v5"><img src="https://img.shields.io/github/v/release/Goldziher/ai-rulez?label=go&color=blue" alt="Go version"></a>
<a href="https://github.com/Goldziher/ai-rulez/blob/main/LICENSE"><img src="https://img.shields.io/github/license/Goldziher/ai-rulez?color=blue" alt="License"></a>
<a href="https://goldziher.github.io/ai-rulez/"><img src="https://img.shields.io/badge/docs-ai--rulez-blue" alt="Documentation"></a>
</p>
<p align="center">
<a href="https://goldziher.github.io/ai-rulez/"><strong>Documentation</strong></a> ·
<a href="https://goldziher.github.io/ai-rulez/quick-start/"><strong>Quick Start</strong></a> ·
<a href="https://goldziher.github.io/ai-rulez/examples/"><strong>Examples</strong></a>
</p>
---
## What it is
ai-rulez is a standards-compliant lifecycle tool for agent knowledge and capabilities. You keep rules, context, skills,
agents, commands, hooks, permissions and MCP servers in one source of truth, `.ai-rulez/`, and ai-rulez takes them
through the whole lifecycle:
```text
author -> generate -> bundle -> lint / validate -> govern -> publish
```
Most tools in this space stop at "turn my rules into markdown." ai-rulez treats agent knowledge like software you
ship: versioned, linted, locked, signed and audited, with an organization policy layer that enforces a floor no
repository can loosen.
- **One source, 52 targets.** Author once in `.ai-rulez/`; generate native files for every harness your team uses instead of hand-copying rules across tools.
- **A real lock file.** `ai-rulez.lock` pins remote includes, installed skills and authored content the way `go.sum` or `package-lock.json` pin dependencies.
- **Supply-chain-grade trust.** Sigstore signing, in-toto/DSSE attestations, CycloneDX/SPDX SBOMs and reviewer approvals, so "what prompt is actually running in production" has a verifiable answer.
- **Org-wide policy, not just repo config.** A central [policy](docs/policy.md) sets tighten-only floors; `verify` and CI gates enforce it everywhere.
## Features
The table covers the implemented capabilities. The [CLI reference](docs/cli-reference.md) lists every command and flag;
[standards](#standards) below distinguishes supported formats from partial integrations.
| Feature | What it does | Commands / configuration | Docs |
| ------- | ------------ | ------------------------ | ---- |
| Content authoring | Manage rules, context, skills with scripts and references, agents, commands and code-review checks. | `init`, `add`, `edit`, `show`, `list`, `remove` | [Configuration](docs/configuration.md), [Skills](docs/skills.md), [Checks](docs/checks.md) |
| Open-standard sources | Use `.ai-rulez/` as an OKF bundle, or author in `AGENTS.md` and `.agents/` without generation overwriting that source. Convert a pre-OKF content tree in place. | `source = "agents"`, `migrate okf`, `export okf`, `import okf`, `okf validate` | [OKF](docs/okf.md), [Agents source](docs/agents-source.md) |
| Domains, profiles and roles | Organize content by domain, select domains by profile, and compose roles that filter items and control skill modes. | `domain`, `profile`, `roles`, `--profile`, `--role` | [Domains and profiles](docs/domains.md), [Roles](docs/roles.md) |
| Configuration layers and local overrides | Merge named TOML layers, a developer's local overlay and user-level settings; inspect the effective local changes. | `config layers`, `local`, `config.<name>.toml`, `--no-local` | [Layers](docs/configuration.md#layered-configuration), [Local configuration](docs/local-overrides.md) |
| Shared includes and skill installation | Reuse local or Git content, choose merge precedence, install individual skills and load skill sources with pinned revisions or version ranges. | `include`, `skill install/list/remove/update`, `includes`, `installed_skills`, `skill_sources` | [Includes](docs/includes.md), [Installed skills](docs/installed-skills.md) |
| Builtin content | Select bundled domains and individual guidance items, or exclude items from a selected pack. | `builtins list/show`, `builtins` | [Builtin domains](docs/domains.md) |
| Import and migration | Convert existing harness files and supported third-party layouts with a lossiness report; migrate a 4.x configuration to v5. | `convert`, `migrate v5` | [Conversion](docs/cli.md#convert-command), [Migration](docs/migration-v5.md) |
| Native generation | Render the file layouts, frontmatter and settings of 52 built-in harnesses, including Claude Code, Cursor, Codex, Copilot and Gemini CLI. | `generate`, `presets` | [Harness matrix](docs/harnesses.md) |
| Custom presets | Define templates or provider-backed presets for an additional harness and its supported content and settings. | `presets` inline tables, provider specs | [Custom presets](docs/profiles.md), [Provider-backed presets](docs/configuration.md#provider-backed-presets-full-parity) |
| Hooks, permissions and MCP settings | Translate shared hooks, allow/deny rules and MCP server definitions to each harness's supported native settings; report unsupported translations. | `hooks`, `permissions`, `mcp_servers` | [Settings](docs/settings.md), [Permissions](docs/permissions.md) |
| Remote-session setup | A `SessionStart` hook (on by default for Claude Code) that runs `ai-rulez generate` when a cloud session starts. | `[session_start]`, `session-start` | [Remote sessions](docs/session-start.md) |
| Generated-file guard | Install a supported harness's PreToolUse hook to block agent edits to generated files. | `[guard] generated = true`, `guard` | [Guard](docs/cli.md#guard-command) |
| User scope and monorepos | Generate user-level assistant configuration, watch source changes, or process nested project roots with ordered reports. | `generate --user`, `generate --watch`, `generate/validate --recursive` | [User scope](docs/user-scope.md), [Monorepos](docs/monorepo.md) |
| Drift, cleanup and diagnostics | Preview a generation plan, detect output drift, remove generated files and diagnose configuration, locks, tools and MCP placeholders. | `generate --dry-run/--emit-plan/--check`, `clean`, `doctor` | [CLI reference](docs/cli.md), [Architecture](docs/architecture.md) |
| Plugin and marketplace authoring | Bundle runtime-specific plugins and marketplace indexes, including multi-plugin definitions, Agent Plugins and opt-in Pi packages. | `generate --plugin`, `plugin`, `plugins_authoring`, `marketplaces_authoring` | [Plugins](docs/plugins.md), [Agent Plugins](docs/agent-plugins.md) |
| Discovery and documentation formats | Generate `AGENTS.md` and `llms.txt`; emit ARD manifests and experimental MCP Server Cards. | `agents_md`, `llms-txt` preset, `publish --emit`, `publish emit` | [AGENTS.md](docs/agents-md.md), [llms.txt](docs/llms-txt.md), [ARD](docs/ard.md), [MCP Server Cards](docs/mcp-server-card.md) |
| Release artifacts and publishing | Build reproducible archives, manifests and checksums, optionally sign them, verify a dist directory and explicitly upload to GitHub releases, npm or OCI. | `publish`, `publish verify`, `--to`, `--execute --yes` | [Publishing](docs/publish.md) |
| Static validation and security scanning | Check config, content, scripts and conventional harness files with stable `AR` codes, without requiring an ai-rulez config. Scope by paths, globs or Git changes; use fixes, suppressions, baselines and JSON/SARIF reports. | `validate`, `scan`, `--strict`, `--path`, `--glob`, `--outputs`, `--changed` | [Validation and security rules](docs/strict-validation.md), [Harness traps](docs/harness-traps.md) |
| External scanners | Run configured scanner programs with explicit execution/egress controls and merge their findings into validation reports. | `lint.external`, `scanners list/doctor`, `validate/scan --external` | [Scanner integrations](docs/strict-validation.md) |
| Repository verifiers | Run declared deterministic checks, opt into command or LLM predicates, link checks to rules, test fixtures and calibrate model-backed checks. | `verifiers run/list/explain/test/calibrate/suggest`, `validate --verifiers` | [Verifiers](docs/verifiers.md) |
| Model-based judge | Review agent configuration for security, quality and governance with quoted evidence, an adjudicator, rubric skills, change-impact selection and explicit incomplete coverage. Estimate before sending; cap spend and emit JSON or SARIF. | `judge`, `--estimate`, `--allow-llm`, `--changed`, `--skill`, `--fail-on` | [Judge](docs/judge.md) |
| Rubric review and patch proposals | Score configured items offline or with a semantic model, calibrate a rubric's gate and propose verified patches for findings. | `review`, `review calibrate/explain/fix`, `rubric` | [Review](docs/review.md) |
| Skill evaluation | Run harness-neutral eval cases through adapters; measure pass rate, trigger precision/recall, ablation and cost; import scenarios and calibrate estimates. | `eval run/import/calibrate-estimate` | [Evals](docs/evals.md) |
| Skill improvement (experimental) | Run an external optimizer on a copy, gate candidates against held-out evals, inspect results, apply an accepted change or open an isolated PR. | `improve run/show/apply/pr/clean/adapters` | [Improve](docs/improve.md) |
| Locks and source updates | Pin remote sources, installed skills, authored content and outputs; check drift, inspect outdated sources and update version-range pins. | `lock`, `lock --check/--diff/--outdated`, `update`, `skill update` | [Lock file](docs/lockfile.md) |
| Organization policy | Apply central floors for sources, approvals, signatures, execution and model use that repository settings cannot loosen. | `--policy`, `AI_RULEZ_POLICY`, `validate --show-policy` | [Policy](docs/policy.md), [Trust model](docs/trust-model.md) |
| Reviewer approvals | Bind approvals to content digests, review diffs, revoke or deny content, enforce authorized reviewers and verify signed approvals. | `approve`, `verify --approvals` | [Approvals](docs/approvals.md) |
| Signatures and attestations | Sign locks, bundles, skills, SBOMs and policies; verify Sigstore bundles, in-toto/DSSE attestations and publisher provenance. | `sign`, `verify --attestation/--self/--plugin`, `trust update` | [Signing](docs/signing.md) |
| Software bills of materials | Inventory AI content, dependencies and MCP packages in CycloneDX 1.6 or SPDX 2.3; detect SBOM drift and enforce lock and pin gates. | `sbom`, `sbom --check -o <file>`, `--require-lock`, `--strict-pins`, `--type spdx-json` | [SBOM](docs/sbom.md) |
| MCP management and skill serving | Expose configuration tools and reports to assistants; serve skills and supporting resources on demand with role filters and session budgets. Select a read-only server or reuse a local daemon. | `mcp`, `--serve-skills`, `--read-only`, `--daemon` | [MCP server](docs/mcp-server.md) |
| Skill search | Rank skills lexically, by vectors or with hybrid embeddings, maintain the index, evaluate ranking and mine queries from usage. | `search`, `search index/status/mine`, `search --eval` | [Search](docs/search.md) |
| Catalog and inventory | Inspect item metadata, owners, roles, dependencies and lock status; join eval/usage results, compare catalogs and build a static catalog website. | `catalog`, `catalog diff`, `catalog --html` | [Catalog](docs/catalog.md) |
| Token and context budgets | Report generated prompt tokens, compare profiles and roles, identify costly content and gate always-loaded or on-demand budgets. | `tokens`, `cost`, `--budget`, `--on-demand-budget` | [Token reports](docs/cli.md#tokens-command), [Cost reports](docs/cli.md#cost-command) |
| Usage and item-load telemetry | Opt into identifier-only skill invocation, feedback and item-load records; report usage and optionally export item-load telemetry through OTLP. | `telemetry record/report/feedback/enable/export` | [Usage telemetry](docs/usage-telemetry.md), [Item loads and OTLP](docs/telemetry.md) |
| Shared model access | Route model calls through liter-llm, including AWS Bedrock credential refresh and OpenAI-compatible gateways, with network consent, budgets, authenticated caching and redacted diagnostics. | `[llm]`, `llm doctor`, `llm estimate` | [LLM access](docs/llm.md) |
| Automation and embedding | Use the composite GitHub Action, pre-commit/prek, Poly or Lefthook; call the experimental Go API to load, validate, plan/apply generation and manage locks. Shell completion and JSON schemas support other integrations. | [Action](action.yml), `init --setup-hooks`, `completion`, `pkg/airulez` | [Git hooks](#git-hooks), [GitHub Actions](#github-actions), [Go API](docs/embedding.md), [Schemas](docs/schema.md) |
<p align="center">
<img src="https://raw.githubusercontent.com/Goldziher/ai-rulez/main/docs/assets/quickstart.gif" alt="ai-rulez init and generate: one .ai-rulez source becomes the native files of four AI tools" width="820" />
</p>
## Supported tools
ai-rulez ships 52 built-in presets — one per AI coding assistant, CLI agent or IDE integration — each writing that
tool's own file layout, frontmatter and settings documents from the same `.ai-rulez/` source. Mix and match any
combination in `presets = [...]`; the full feature matrix (rules, skills, agents, commands, MCP, hooks, permissions,
checks, user scope) is in [Supported harnesses](docs/harnesses.md), and a provider that isn't listed yet still gets
full parity through a [custom provider-backed preset](docs/configuration.md#provider-backed-presets-full-parity).
<details>
<summary><strong>All 52 supported tools</strong></summary>
| Preset | Tool |
| ------ | ---- |
| `aiassistant` | [JetBrains AI Assistant](https://www.jetbrains.com/help/ai-assistant/) |
| `amp` | [Sourcegraph Amp](https://ampcode.com) |
| `antigravity` | [Google Antigravity](https://antigravity.google/docs) |
| `augment` | [Augment Code](https://docs.augmentcode.com) |
| `baz` | [Baz](https://baz.ai/docs/agents/skills-and-instructions) |
| `bob` | [IBM Bob](https://bob.ibm.com/docs) |
| `claude` | [Claude Code](https://code.claude.com/docs) |
| `cline` | [Cline](https://docs.cline.bot) |
| `codebuddy` | [CodeBuddy Code](https://www.codebuddy.ai/docs/cli) |
| `codebuff` | [Codebuff](https://www.codebuff.com/docs) |
| `codewhale` | [Codewhale](https://github.com/codewhale-hq/Codewhale) |
| `codex` | [OpenAI Codex](https://developers.openai.com/codex) |
| `commandcode` | [Command Code](https://commandcode.ai/docs) |
| `copilot` | [GitHub Copilot](https://docs.github.com/en/copilot) |
| `copilot-cli` | [GitHub Copilot CLI](https://docs.github.com/en/copilot/how-tos/copilot-cli) |
| `cortex` | [Snowflake Cortex Code](https://docs.snowflake.com/en/user-guide/cortex-code/extensibility) |
| `crush` | [Crush](https://github.com/charmbracelet/crush) |
| `cursor` | [Cursor](https://cursor.com/docs) |
| `deepagents` | [Deep Agents Code](https://docs.langchain.com/oss/deepagents/code) |
| `devin` | [Devin CLI](https://docs.devin.ai/cli) |
| `dsh` | [DeepSeek Harness](https://github.com/deepseek-ai/deepseek-harness) |
| `factory` | [Factory Droid](https://docs.factory.ai) |
| `gemini` | [Gemini CLI](https://geminicli.com/docs/) |
| `gitlab-duo` | [GitLab Duo CLI](https://docs.gitlab.com/user/gitlab_duo_cli/customize/) |
| `goose` | [goose](https://goose-docs.ai) |
| `grok` | [Grok Build CLI](https://docs.x.ai/build/overview) |
| `hermes` | [Hermes Agent](https://hermes-agent.nousresearch.com/docs) |
| `junie` | [JetBrains Junie](https://junie.jetbrains.com) |
| `kilo` | [Kilo Code](https://kilo.ai/docs) |
| `kimi` | [Kimi Code](https://moonshotai.github.io/kimi-code/) |
| `kiro` | [Kiro](https://kiro.dev/docs) |
| `letta` | [Letta Code](https://docs.letta.com) |
| `mimocode` | [MiMo Code](https://mimo.xiaomi.com/mimocode) |
| `muse` | [Meta Muse Code](https://dev.meta.ai/docs/muse-code) |
| `omp` | [oh-my-pi](https://github.com/can1357/oh-my-pi) |
| `openclaw` | [OpenClaw](https://docs.openclaw.ai) |
| `opencode` | [OpenCode](https://opencode.ai/docs) |
| `pi` | [pi](https://pi.dev/docs/latest) |
| `poolside` | [Poolside Pool](https://docs.poolside.ai) |
| `qoder` | [Qoder](https://docs.qoder.com/cli/quickstart) |
| `qwen` | [Qwen Code](https://qwenlm.github.io/qwen-code-docs/en/) |
| `reasonix` | [Reasonix](https://github.com/esengine/DeepSeek-Reasonix) |
| `replit` | [Replit Agent](https://docs.replit.com/features/agent/overview) |
| `rovodev` | [Atlassian Rovo Dev](https://www.atlassian.com/software/rovo-dev) |
| `takt` | [Takt](https://github.com/nrslib/takt) |
| `trae` | [Trae](https://docs.trae.ai/ide/rules) |
| `vibe` | [Mistral Vibe](https://docs.mistral.ai/vibe) |
| `warp` | [Warp](https://docs.warp.dev) |
| `xum` | [Xum](https://xum.coder.com) |
| `zcode` | [ZCode](https://zcode.z.ai/en/docs) |
| `zed` | [Zed](https://zed.dev/docs/ai) |
| `zoocode` | [Zoo Code](https://docs.zoocode.dev) |
</details>
## Standards
Generation isn't limited to harness files — ai-rulez aims to be compliant with the major emerging standards in the agentic ecosystem. A standard is listed when the code generates and validates it today.
[docs/standards.md](docs/standards.md) carries the pinned spec version, the conformance test suite and the known gaps for
each row, and is the source of truth for test status.
| Standard | Status | Generate / bundle | Lint / validate |
| -------- | ------ | ----------------- | --------------- |
| [OKF](docs/okf.md) (Open Knowledge Format, v0.2), the internal format of `.ai-rulez/` | supported | `export okf`, the `okf` preset | `okf validate` |
| [Agent Plugins](docs/agent-plugins.md) (agent-plugins.org) | supported | `generate --plugin`, `publish --emit agent-plugins` | `validate --strict` (AR9O codes) |
| [ARD](docs/ard.md) (Agentic Resource Discovery, a proposal) | supported | `publish --emit ard` | `validate` (entry schema, URN grammar) |
| [Agent Skills](docs/skills.md) (agentskills.io) | supported | `generate` | `validate` (frontmatter checks; partial) |
| [AGENTS.md](docs/agents-md.md), also as the [source](docs/agents-source.md) with `source = "agents"` | supported | `generate` | `validate` |
| [llms.txt](docs/llms-txt.md) | supported | the `llms-txt` preset | `validate` |
| MCP server config | supported | `generate` | `validate` (partial) |
| MCP server card | partial (experimental extension, pinned `526201b`) | `publish --emit mcp-server-card`, `publish --emit ard` | conformance tests only |
| [CycloneDX / SPDX SBOM](docs/sbom.md) | supported | `sbom` | `sbom --check` |
| [in-toto / DSSE / Sigstore](docs/signing.md) | supported | `sign` | `verify --attestation` |
| [OpenTelemetry (OTLP)](docs/telemetry.md) | supported | `telemetry export --to otlp` | not applicable |
## Usage
Every option below produces the same `.ai-rulez/` tree and the same generated output — pick however your project
already runs tools. Each one walks through init, generate and validate; [Bundle and publish](#bundle-and-publish),
[Validate](#validate) and [Govern](#govern) below cover what comes after.
```toml
# .ai-rulez/config.toml
presets = ["claude", "cursor", "codex", "copilot", "gemini"] # any of the 52 harnesses
```
<details>
<summary><strong>npx (Node.js, no install)</strong></summary>
```bash
npx ai-rulez@latest init
npx ai-rulez@latest generate
npx ai-rulez@latest validate --strict
```
</details>
<details>
<summary><strong>uvx (Python, no install)</strong></summary>
```bash
uvx ai-rulez init
uvx ai-rulez generate
uvx ai-rulez validate --strict
```
</details>
<details>
<summary><strong>Homebrew (macOS / Linux)</strong></summary>
```bash
brew install goldziher/tap/ai-rulez
ai-rulez init
ai-rulez generate
ai-rulez validate --strict
```
</details>
<details>
<summary><strong>Go</strong></summary>
```bash
go install github.com/Goldziher/ai-rulez/v5/cmd/ai-rulez@latest
ai-rulez init
ai-rulez generate
ai-rulez validate --strict
```
The module path needs `/v5`; without it `@latest` resolves to an old build.
This local build needs cgo, a C toolchain and the Go module cache at runtime.
For a statically linked build, use `task build` from a clone; see [source installation](docs/installation.md).
</details>
<details>
<summary><strong>npm (global)</strong></summary>
```bash
npm install -g ai-rulez
ai-rulez init
ai-rulez generate
ai-rulez validate --strict
```
</details>
<details>
<summary><strong>uv tool</strong></summary>
```bash
uv tool install ai-rulez
ai-rulez init
ai-rulez generate
ai-rulez validate --strict
```
</details>
<details>
<summary><strong>pip / pipx</strong></summary>
```bash
pip install ai-rulez
# or, isolated:
pipx install ai-rulez
ai-rulez init
ai-rulez generate
ai-rulez validate --strict
```
</details>
<details>
<summary><strong>Day-to-day commands</strong></summary>
```bash
ai-rulez generate --watch # regenerate on every change to .ai-rulez/
ai-rulez generate --check # CI: exit 2 when committed outputs differ from the sources
ai-rulez doctor # read-only diagnostics: drift, removed presets, missing tools
ai-rulez generate --user # the same, for your home directory (~/.claude, ~/.codex, ...)
```
</details>
Prefer the project-level [`.config/` convention](https://github.com/pi0/config-dir)? ai-rulez also discovers
`.config/ai-rulez/`, and `ai-rulez init --config-dir .config/ai-rulez` scaffolds it.
## Bundle and publish
Bundling and publishing need a `[plugin]` block (`name`, `version` and `description`) in `config.toml`:
<details>
<summary><strong>Show commands</strong></summary>
```bash
ai-rulez generate --plugin # plugin bundles and a marketplace index from the [plugin] block
ai-rulez export okf # an OKF bundle of your rules, context and skills (docs/okf)
ai-rulez publish --emit agent-plugins --emit ard --dry-run
ai-rulez publish --to github-release --execute --yes
```
</details>
`publish` runs a preflight first (`validate --strict`, `lock --check`, `verify --plugin`, a secret scan and the policy
gates), writes a byte-reproducible tar.gz with `SHA256SUMS`, and uploads only when you pass `--execute --yes`. See
[Publish](docs/publish.md) and [Authoring plugins](docs/plugins.md).
Pi packages are opt-in with `[plugin] runtimes = ["pi"]`: skills and prompt templates can be installed
from Git or npm. Publishing to npm preserves Pi and OpenCode runtime metadata while using the
configured scope and plugin name for the released package.
`verify --plugin` reports a hand-edited or stray file in a bundle as drift, and `generate --plugin` restores it:
<p align="center">
<img src="https://raw.githubusercontent.com/Goldziher/ai-rulez/main/docs/assets/plugin-drift.gif" alt="ai-rulez verify --plugin reporting changed and extra files in a plugin bundle, then generate --plugin restoring it" width="820" />
</p>
## Validate
<p align="center">
<img src="https://raw.githubusercontent.com/Goldziher/ai-rulez/main/docs/assets/validate.gif" alt="ai-rulez validate --strict flags a defect, --explain describes it, and generate --check fails on drift" width="820" />
</p>
<details>
<summary><strong>Show commands</strong></summary>
```bash
ai-rulez validate # config plus content checks, exit 2 on findings
ai-rulez validate --explain AR001 # what a rule checks and how to suppress it
ai-rulez validate --format sarif --output ai-rulez.sarif
ai-rulez validate --path catalog --glob "plugins/**/*.json" # also lint files outside .ai-rulez/
ai-rulez validate --outputs # also lint the generated files; each problem is reported once
ai-rulez validate --strict # in any repo: with no config it lints AGENTS.md, SKILL.md, .claude/, .agents/
ai-rulez okf validate docs/okf # lint any OKF bundle, ai-rulez's or a third party's
ai-rulez generate --check # generated files still match their sources
```
</details>
## Govern
<details>
<summary><strong>Show commands</strong></summary>
```bash
ai-rulez lock # pin remote includes, installed skills and authored content
ai-rulez approve --list # what still needs a reviewer's approval
ai-rulez sign --lock # a Sigstore bundle (DSSE over an in-toto statement) for the lock
ai-rulez verify --attestation # check it offline against the [signing] policy
ai-rulez sbom -o ai-bom.cdx.json # CycloneDX 1.6 (or --type spdx-json)
```
</details>
An organization [policy](docs/policy.md) sets tighten-only floors a repository cannot loosen. See the
[trust model](docs/trust-model.md).
## Operate
- **MCP server.** `ai-rulez mcp` lets an assistant manage the configuration; with `--serve-skills` one server serves
skills on demand and the config tools together (`--read-only` keeps only the read-only tools).
See [MCP server](docs/mcp-server.md).
Served skills cost nothing until an assistant asks for them:
<img src="https://raw.githubusercontent.com/Goldziher/ai-rulez/main/docs/assets/skills-mcp.gif" alt="ai-rulez tokens before and after serving skills over MCP, then find_skill and load_skill" width="780" />
- **Monorepos.** Nested roots each carry their own `.ai-rulez/`; `generate --recursive` and
`validate --recursive` walk them all, and `tokens --budget` gates what every session loads.
See [Monorepos](docs/monorepo.md).
<img src="https://raw.githubusercontent.com/Goldziher/ai-rulez/main/docs/assets/monorepo.gif" alt="ai-rulez generate --recursive over two nested roots, recursive validation, and a token budget check" width="780" />
- **Telemetry.** Item-load records with an optional OTLP export. See [telemetry](docs/telemetry.md).
- **Evals and improve.** `ai-rulez eval` scores skills; `improve` (experimental) optimizes them behind a held-out eval
gate. See [evals](docs/evals.md) and [improve](docs/improve.md).
- **Judge.** `ai-rulez judge` reviews skills, agents, commands, rules, `CLAUDE.md`, `AGENTS.md`, MCP and settings files
with models, from the command line alone: security, quality and governance reviewers, a verifier that checks quoted
evidence, an independent adjudicator, and scores computed in code. Failed rubric criteria without a verifiable quote
are reported separately as unit-level findings. Governance reads the rules and
`CLAUDE.md` files that apply to each unit, nested `.claude/` directories at any depth. `ai-rulez judge --estimate`, then
`--allow-llm` to send; `--changed=origin/main` for a pull request, `--format sarif` for code scanning. See [judge](docs/judge.md).
- **Cost and inventory.** `tokens`, `cost`, `catalog` and `search` report what the configuration contains and costs.
Existing project on 4.x? Run `ai-rulez migrate v5 --dry-run`, then see [Migrating to v5](docs/migration-v5.md).
## Documentation
Full documentation at [goldziher.github.io/ai-rulez](https://goldziher.github.io/ai-rulez/). For how a source becomes
generated files, and the package map, see [Architecture](docs/architecture.md).
## Git Hooks
ai-rulez supports multiple pre-commit hook harnesses.
<details>
<summary><strong>pre-commit/prek</strong></summary>
Add to `.pre-commit-config.yaml`:
```yaml
repos:
- repo: https://github.com/Goldziher/ai-rulez
rev: v5.6.2
hooks:
- id: ai-rulez-recursive # generate: write outputs for every nested .ai-rulez/ root
- id: ai-rulez-validate # validate --recursive: content, security and link checks
```
Every hook runs `scripts/pre-commit/run-ai-rulez.sh` and triggers only on root or nested `.ai-rulez/` changes:
| Hook id | Runs |
| --- | --- |
| `ai-rulez-generate` | `generate`, the root `.ai-rulez/` only |
| `ai-rulez-recursive` | `generate --recursive`, every nested root |
| `ai-rulez-validate` | `validate --recursive` — a real validation run, not a dry run |
| `ai-rulez-plugin-generate` | `generate --recursive --plugin --if-configured` |
| `ai-rulez-plugin-verify` | `verify --recursive --plugin --if-configured` |
Add `ai-rulez-plugin-generate` and `ai-rulez-plugin-verify` too if the repository ships plugin bundles; the
`--if-configured` flag makes them no-ops on consumer-only repositories.
</details>
<details>
<summary><strong>Poly</strong></summary>
Add ai-rulez as a managed source in your existing `poly.toml` and select the hooks your
repository needs:
```toml
[[hooks.sources]]
id = "ai-rulez"
git = "https://github.com/Goldziher/ai-rulez.git"
revision = "v5.6.2"
hooks = ["ai-rulez-recursive", "ai-rulez-plugin-verify"]
```
The source also provides `ai-rulez-validate`, `ai-rulez-generate`,
and `ai-rulez-plugin-generate`.
Plugin hooks use `--if-configured`, so they skip consumer-only repositories that
do not contain a producer `[plugin]` or multi-member `[marketplace]` block.
Resolve and commit the source lock, then install the Git shims:
```bash
poly hooks update
git add poly.toml poly-hooks.lock
poly hooks install
```
</details>
<details>
<summary><strong>Lefthook</strong></summary>
Add to `lefthook.yml`:
```yaml
pre-commit:
commands:
ai-rulez:
glob: ".ai-rulez/**"
run: ai-rulez generate --recursive --no-local
```
Hooks run on each developer's machine and would otherwise load that developer's gitignored
`config.local.toml` overlay and `.ai-rulez/local/` content. Pass `--no-local` (also accepted by `validate`
and `tokens`) where a hook should see only the shared configuration, as a teammate or CI does. If you do
not, `generate` still refuses to write local-derived changes into tracked shared files; never add
`--allow-local-drift` to a hook. See [Poly hooks guide](docs/poly-hooks.md#machine-local-configuration-in-hooks).
In a monorepo, `ai-rulez generate --recursive` and `ai-rulez validate --recursive` process every nested
`.ai-rulez/` root, report all failures, and exit non-zero if any root failed.
Or run `ai-rulez init --setup-hooks` while initializing a repo to wire hooks in automatically.
</details>
## GitHub Actions
ai-rulez ships its own composite Action ([`action.yml`](action.yml) at the repository root), so any workflow can
`uses: Goldziher/ai-rulez@<ref>` directly — no subpath. It downloads a release binary, verifies its sha256 against
the release's `checksums.txt`, caches it (keyed by OS, arch and the resolved version) and adds it to `PATH`. No Go
toolchain, Node or Python needed in the job. It reuses the same `scripts/pre-commit/run-ai-rulez.sh`
download-and-verify logic the pre-commit hooks above use, so the two stay in sync.
Use `Goldziher/ai-rulez@v5` for compatible action updates: publishing a stable v5 release advances
this floating tag after all distribution jobs succeed. Pin a full release tag or commit SHA to fix
the action code. The action ref and installed CLI version are independent: set `version` as well
to pin the binary; its default `latest` can eventually select a newer major release.
| | |
| --- | --- |
| **Input** `version` | `X.Y.Z`, `vX.Y.Z`, or `latest` (default). Optional — omit it to always use the newest release. |
| **Inputs** `llm-*` | Optional provider, model, gateway, credential variable name, network opt-in, budgets and other LLM settings exported to later steps. [Full reference](docs/github-actions.md#llm-configuration-inputs). |
| **Output** `version` | The resolved version, without the leading `v`. |
| **Output** `path` | Absolute path to the installed `ai-rulez` binary. |
<details>
<summary><strong>Show workflow</strong></summary>
```yaml
name: ai-rulez
on:
pull_request:
push:
branches: [main]
jobs:
check:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- uses: Goldziher/ai-rulez@v5
with:
version: 5.6.2 # pin the CLI independently of the floating action ref
- run: ai-rulez generate --check # the committed output matches the sources
- run: ai-rulez validate --strict # content and security checks, warnings fail
- run: ai-rulez lock --check # pins were updated when the sources were
```
</details>
For model-backed reviews, configure `llm-model`, `llm-allow-network: 'true'` and spend limits,
pass provider credentials through `env:` on the command step or job, and run `ai-rulez judge
--allow-llm`. Installation itself makes no model requests. See the [action guide](docs/github-actions.md)
for all inputs, environment precedence, Bedrock, outputs and complete examples.
See [CI usage](docs/lockfile.md#ci-usage) for the lock-check-vs-generate-check distinction and a scheduled
"outdated sources" report job.