{
"markdown": "# AgentsKit Registry\n\nProfile: top-level-repository\n\n[](https://github.com/AgentsKit-io/agentskit-registry/actions/workflows/ci.yml)\n[](https://registry.agentskit.io/agents)\n[](./LICENSE)\n[](https://www.bestpractices.dev/projects/13876)\n\n**Copy a production-minded AI agent into your project and own the source.**\nAgentsKit Registry is a shadcn-style catalog of 346 validated agents built with\n[AgentsKit](https://www.agentskit.io/docs). It adds no Registry runtime and no\nlock-in: the CLI copies readable TypeScript that your team can review and edit.\n\nIt is intended for teams that want a ready agent starting point without adopting\na proprietary registry runtime.\n\n**Tags:** `agentskit` · `agent-registry` · `typescript` · `ai-agents` · `shadcn-style`\n\n**Topics:** `ai-agents` · `registry` · `developer-experience`\n\n[Browse the Registry](https://registry.agentskit.io/agents) ·\n[Start in five minutes](./docs/getting-started.md) ·\n[Contribute an agent](./CONTRIBUTING.md) ·\n[Security](./SECURITY.md) ·\n[Governance](./GOVERNANCE.md) ·\n[Code of Conduct](./CODE_OF_CONDUCT.md)\n\n\n\n## Verified proof\n\n- Catalog counts come from [`public/r/index.json`](./public/r/index.json) and [`ecosystem-claims.json`](./ecosystem-claims.json).\n- Clean-fixture install/run contract: `npm run test:quickstart` (`scripts/quickstart.test.mjs`).\n- Structural validation and Doc Bridge gates: `npm run validate`, `npm run docs:bridge:gate`.\n- Deterministic discovery build: `npm run discovery:build`.\n- Credential-free catalog proof: `node examples/verify-readme.mjs`.\n\n## Validate agents in GitHub Actions\n\nUse the Registry contract as a CI gate for copy-owned agent folders in another\nrepository. The action checks required source, test, metadata, and README files;\nmetadata IDs and referenced files; and the Registry content policy. It does not\nsend source code to an API or require credentials.\n\n```yaml\nname: Validate agents\n\non:\n pull_request:\n push:\n branches: [main]\n\njobs:\n agentskit:\n runs-on: ubuntu-latest\n steps:\n - uses: actions/checkout@v4\n - uses: AgentsKit-io/agentskit-registry@v1\n with:\n path: agents\n```\n\nThe `path` input may point to one agent folder or to a directory whose immediate\nchildren are agent folders.\n\n\n> Registry is **beta**. All 346 entries pass structural and repository tests;\n> 333 also publish sanitized independent-review evidence. Templates remain code\n> you must review against your provider, tools, data policy, and risk profile.\n\n## From catalog to code\n\n```mermaid\nflowchart LR\n F[\"Find by ID, category, or capability\"] --> A[\"npx agentskit add \"]\n A --> C[\"Readable source copied locally\"]\n C --> R[\"Review and adapt\"]\n R --> S[\"Ship with your adapter and policy\"]\n```\n\n\n```bash\nnpx agentskit add research\n```\n\n```ts\nimport { openai } from '@agentskit/adapters'\nimport { createResearchAgent } from './agents/research/agent'\n\nconst agent = createResearchAgent({\n adapter: openai({\n apiKey: process.env.OPENAI_API_KEY!,\n model: 'gpt-4o',\n }),\n})\n\nconst result = await agent.run('What changed in the EU AI Act?')\nconsole.log(result.content)\n```\n\nSwap the adapter, tools, memory, model, and policy through supported AgentsKit\ncontracts. The copied factory stays ordinary application code.\n\n## What is actually here\n\n- **346 validated, installable agents** across 23 categories, from coding and\n security to legal, clinical, marketing, operations, support, and research.\n- **333 independent-review summaries** with scores, confidence, strengths, and\n required adjustments; private raw reviewer output is never published.\n- **Typed source + focused tests + usage guide** for every installable entry.\n- **Eval replay fixtures** for repeatable behavioral checks without live keys.\n- **JSON, `llms.txt`, MCP, and deterministic discovery artifacts** for people,\n tools, and LLMs to navigate the same canonical catalog.\n\nClaims are derived from [`public/r/index.json`](./public/r/index.json) and the\ngenerated [`ecosystem-claims.json`](./ecosystem-claims.json), not hand-maintained\nmarketing copy.\n\n## Discovery without unnecessary backend calls\n\nExact agent IDs, titles, install commands, categories, capabilities,\ncontribution links, and ecosystem navigation resolve from a signed local\nartifact. Semantic requests such as “which agent fits my incident workflow?”\nescalate through AgentsKit Chat to the trusted cited backend only after a\nvalidated local miss.\n\nSee [architecture and ownership](./docs/architecture.md) for the full boundary.\n\n## Repository ownership\n\nThis repository owns catalog data and generated machine artifacts. The live\nFumadocs application and its AgentsChat presentation live in\n[`AgentsKit-io/agentskit/apps/registry`](https://github.com/AgentsKit-io/agentskit/tree/main/apps/registry).\nAgentsKit owns runtime primitives and the CLI; AgentsKit Chat owns the shared\nanswer protocol and renderer behavior. There is no second framework here.\n\n## Contributing\n\nNew agents and improvements are welcome. Follow [CONTRIBUTING.md](./CONTRIBUTING.md),\nuse a neighboring validated agent as a reference, and run:\n\n```bash\nnpm run validate\nnpm run lint\nnpm test\nnpm run eval:run -- --ecosystem-doc-bridge\nnpm run build\nnpm run docs:bridge:gate\n```\n\nGenerated indexes, claims, deterministic artifacts, and Doc Bridge handoffs\nmust remain fresh in the pull request.\n\n## Maturity\n\nRegistry is **beta**. All 346 entries pass structural and repository tests; 333\nalso publish sanitized independent-review evidence. Templates remain code you\nmust review against your provider, tools, data policy, and risk profile.\n\n## AgentsKit ecosystem\n\n- [AgentsKit](https://www.agentskit.io/docs) — build the agent runtime.\n- **Registry** — start from owned, ready-made source.\n- [AgentsKit Chat](https://chat.agentskit.io) — deliver one\n agent experience across Web, native, and terminal interfaces.\n- [Agents Playbook](https://playbook.agentskit.io/docs) — apply production\n engineering and review discipline.\n- [Doc Bridge](https://doc-bridge.agentskit.io) — turn documentation\n into executable agent handoffs.\n- [Code Review](https://github.com/AgentsKit-io/code-review-cli) — run deep,\n low-noise review with the model already in use.\n- [AgentsKit OS](https://akos.agentskit.io/docs) — operate and govern agents.\n\n## Compatibility and license\n\nRegistry source targets the package ranges declared by each `meta.json`; the\nrepository CI currently runs on Node.js 22. Review an agent's generated bundle\nbefore upgrading its dependencies. Licensed under [MIT](./LICENSE).\n",
"bytes": 7091,
"sha": "8b83049968582bd6557cf144a002bd75479cb5d0dbf82c77c923bfd44754310a",
"repo_slug": "agentskit-io/agentskit-registry",
"fonte": "repo",
"truncated": false,
"api": "https://agentalog.com/api/listings/mcp_io_github_agentskit_io_agentskit_registr_9c70a1f2/readme"
}