https://github.com/dligthart/wordclaw
AI-native headless CMS - manage content through REST, GraphQL, or Model Context Protocol. Built for agents and humans alike, with versioning, dry-run mode, webhooks, and Lightning micropayments.
https://github.com/dligthart/wordclaw
ai-agents graphql headless-cms l402 rest-api
Last synced: 3 months ago
JSON representation
AI-native headless CMS - manage content through REST, GraphQL, or Model Context Protocol. Built for agents and humans alike, with versioning, dry-run mode, webhooks, and Lightning micropayments.
- Host: GitHub
- URL: https://github.com/dligthart/wordclaw
- Owner: dligthart
- License: mit
- Created: 2026-02-17T13:40:43.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2026-04-02T19:15:41.000Z (3 months ago)
- Last Synced: 2026-04-03T06:46:27.905Z (3 months ago)
- Topics: ai-agents, graphql, headless-cms, l402, rest-api
- Language: TypeScript
- Homepage:
- Size: 8.6 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 9
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
Awesome Lists containing this project
README
# WordClaw
[](https://github.com/dligthart/wordclaw/actions/workflows/test.yml)
[](https://github.com/dligthart/wordclaw/actions/workflows/test.yml)
WordClaw is a safe content runtime for AI agents and human supervisors.
It combines structured content contracts, review-aware workflows, dry-run safety, auditability, paid-content flows, and agent-native REST/MCP access so autonomous systems can work with content without losing governance.
For active product direction, see the [roadmap](doc/reference/roadmap.md) and [RFC index](doc/rfc/index.md).
## What It Includes
- Structured content with JSON Schema validation, globals, localization, versioning, rollback, and published-vs-working-copy reads.
- Governance primitives such as workflows, dry-run mutation paths, audit logs, idempotency, and tenant isolation.
- Schema-aware assets, reusable forms, background jobs, preview tokens, and reverse-reference inspection.
- REST-first and MCP-first agent surfaces, plus a supervisor UI for oversight.
- Optional semantic search via pgvector and embeddings when `OPENAI_API_KEY` is configured.
For the fuller capability breakdown, see [Features](doc/concepts/features.md).
## Quick Start
```bash
git clone https://github.com/dligthart/wordclaw.git
cd wordclaw
npm install
cp .env.example .env
docker compose up -d
npm run db:migrate
npm run dev:all
```
That gives you:
- API: `http://localhost:4000/api`
- Swagger/OpenAPI docs: `http://localhost:4000/documentation`
- MCP endpoint: `http://localhost:4000/mcp`
- Supervisor UI (dev): `http://localhost:5173/ui/`
Verify the runtime:
```bash
curl http://localhost:4000/health
curl http://localhost:4000/api/deployment-status
npx tsx src/cli/index.ts capabilities status
```
If `deployment-status` reports `domainCount: 0`, bootstrap the first domain before creating content types or content items. Prefer the dedicated CLI path locally:
```bash
npx tsx src/cli/index.ts domains create \
--name "Local Development" \
--hostname local.development
```
Equivalent bootstrap paths are MCP `create_domain` and REST `POST /api/domains`. `AUTH_REQUIRED=false` only opens public discovery; write-capable routes still need a credential unless `ALLOW_INSECURE_LOCAL_ADMIN=true` is active in a non-production environment.
The step-by-step version lives in [Getting Started](doc/tutorials/getting-started.md).
## First Commands
```bash
npx tsx src/cli/index.ts capabilities status
npx tsx src/cli/index.ts mcp call guide_task --json '{"taskId":"discover-deployment"}'
npx tsx src/cli/index.ts domains list
npx tsx src/cli/index.ts mcp inspect --mcp-transport http --api-key writer
npx tsx src/cli/index.ts workspace guide
npx tsx src/cli/index.ts content guide
npx tsx src/cli/index.ts content guide --content-type-id 1
```
Use `content guide` without `--content-type-id` when you need to bootstrap a new schema first. It now returns starter guidance for agent memory, task-log, and checkpoint-style content models.
Live verification helpers now live under `scripts/verification/` and are exposed as `npm run verify:*` commands.
## Documentation
### Start Here
- [Getting Started](doc/tutorials/getting-started.md)
- [Runtime Configuration](doc/reference/runtime-configuration.md)
- [Docker Deployment](doc/guides/docker-deployment.md)
### Agent and Operator Surfaces
- [CLI Guide](doc/guides/cli-guide.md)
- [Claude/OpenClaw Memory + RAG](doc/guides/claude-openclaw-memory-rag.md)
- [MCP Integration](doc/guides/mcp-integration.md)
- [API Reference](doc/reference/api-reference.md)
- [LangGraph MCP Starter](doc/guides/langgraph-mcp-starter.md)
### Product and Architecture
- [Features](doc/concepts/features.md)
- [Roadmap](doc/reference/roadmap.md)
- [Architecture](doc/reference/architecture.md)
- [Data Model](doc/reference/data-model.md)
- [Native Vector RAG Guide](doc/guides/native-vector-rag.md)
### Examples and History
- [Demos](doc/guides/demos.md)
- [RFC Index](doc/rfc/index.md)
## Contributing
1. Fork the repository.
2. For major features or architecture changes, start with an RFC in `doc/rfc/`.
3. Create a feature branch.
4. Commit your changes.
5. Open a pull request.
## License
[MIT](LICENSE)