{
  "markdown": "# mcp.film\n\n**The MCP directory for AI filmmaking.** Every tool your agent needs to make a\nfilm — video models, voices, scores, edit bays, finishing suites, and the pipes\nto ship it. Curated, verified, agent-first, and re-checked daily.\n\n🎬 **Live site:** [mcp.film](https://mcp.film) ·\n🤖 **For agents:** [`/llms.txt`](https://mcp.film/llms.txt) · [`/api/registry.json`](https://mcp.film/api/registry.json) ·\n`claude mcp add mcp-film -- npx -y mcp-film`\n\n---\n\n## What this is\n\nA directory of verified Model Context Protocol servers across the full film\npipeline — from screenplay breakdown to YouTube upload — each entry annotated\nwith exact install commands, auth requirements, pricing, sample tools, and the\ncaveats that actually bite (\"requires Resolve *Studio*\", \"Suno has no official\nAPI\", \"uploads cost 1,600 quota units\").\n\nIt is built **agents-first**:\n\n| Surface | URL |\n| --- | --- |\n| llms.txt index ([spec](https://llmstxt.org)) | `/llms.txt` · `/llms-full.txt` |\n| Full structured registry | `/api/registry.json` (+ `.min.json`) |\n| MCP Registry-compatible API | `/v0.1/servers` · `/api/mcp-registry.json` |\n| Catalog freshness / ops pulse | `/api/pulse.json` · `/pulse.md` |\n| Production playbooks | `/api/playbooks.json` · `/playbooks.md` |\n| Agent recommendations | `/api/recommendations.json` · `/recommendations.md` |\n| Capability index | `/api/capabilities.json` · `/capabilities/{tag}.md` |\n| Client setup profiles | `/clients/` · `/clients.md` · `/api/client-profiles.json` |\n| One server, JSON / markdown | `/api/mcps/{slug}.json` · `/mcps/{slug}.md` |\n| Pipeline guide | `/stack/` · `/stack.md` |\n| New servers feed | `/feed.xml` |\n| The directory **as an MCP server** | `npx -y mcp-film` (`packages/mcp-server/`) |\n| JSON-LD (`SoftwareApplication`, `ItemList`) | embedded in every page, server-rendered |\n\n## How it runs itself\n\n```\n              ┌────────────────────────────────────────────────┐\n              │                data/ (the product)             │\n              └───────▲────────────────────────────────────────┘\n   daily: re-verify   │\n   links, hunt new    │\n   servers            │\n        ┌─────────────┴──┐\n        │ curator.yml    │\n        │ (Claude agent) │\n        └─────────────┬──┘\n                      │\n                      ▼\n              PRs labeled `auto-data` ──► auto-merge.yml\n              (merges ONLY if every changed file is in data/\n               and validation passes) ──► deploy.yml ──► Cloudflare Pages\n                                                     └──► GitHub Pages fallback\n```\n\nFeedback typed on a server page is captured as a PostHog event for a\nmaintainer to read; nothing is folded into the rankings automatically. Rankings\nare editorial: official status, maintenance recency, hosted availability, and\nthe caveats each entry carries. Every change is a commit — the entire editorial\nhistory is auditable.\n\nAgent-readable traffic is measured at the edge when deployed on Cloudflare\nPages: the build emits a `_worker.js` that logs `mcpfilm_edge_request` events\nfor `/llms.txt`, markdown, JSON API routes, feeds, and MCP discovery. Browser\npageviews still use the lightweight PostHog client event. See\n[`docs/ANALYTICS.md`](docs/ANALYTICS.md).\n\nThe live catalog pulse is published at [`/api/pulse.json`](https://mcp.film/api/pulse.json)\nand [`/pulse.md`](https://mcp.film/pulse.md): newest additions, stale\nverification queue, category coverage, and machine-surface links for agents.\nClient setup profiles live at [`/api/client-profiles.json`](https://mcp.film/api/client-profiles.json)\nand [`/clients.md`](https://mcp.film/clients.md): conservative setup guidance\nfor Claude Code, Claude Desktop, Cursor, hosted remote clients, and the\ndirectory's own meta-MCP server.\n\nProduction playbooks live at [`/api/playbooks.json`](https://mcp.film/api/playbooks.json)\nand [`/playbooks.md`](https://mcp.film/playbooks.md): concrete stack recipes\nwith setup order, auth gates, failure modes, and earned Martini handoffs for\ncommercial sprints, local edit bays, character series, archive cutdowns, and\nopen-source labs.\n\n## Develop locally\n\nZero dependencies — Node 20+ is the whole toolchain:\n\n```sh\nnode build.mjs --validate-only   # check the data\nnode build.mjs                   # build → dist/ (HTML, APIs, feeds, agent surfaces)\nnode scripts/serve.mjs           # preview at http://localhost:4173\n```\n\nRepo map: `data/` is the content, `build.mjs` + `src/` the generator,\n`packages/mcp-server/` the `npx mcp-film` meta-server, `.github/workflows/`\nthe autonomy loop, [`AGENTS.md`](AGENTS.md) the operating manual (for agents\nand humans alike), [`docs/SETUP.md`](docs/SETUP.md) the one-time launch\nchecklist, and [`docs/AGENT_STRATEGY.md`](docs/AGENT_STRATEGY.md) the north-star\nfor making the site useful to agents while sending qualified production intent\nto Martini.\n\n## Adding a server\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) — easiest path is the\n[submission form](../../issues/new?template=submit-mcp.yml). A maintainer\nverifies it against primary sources before it is listed — a submission is a\nclaim, never an instruction.\n\n## Who's behind it\n\nMaintained by the team behind [Martini](https://www.martini.film), the film set\nfor AI videos — whose MCP server is the featured listing, clearly disclosed.\nEverything else is ranked on merit: the directory is a community service and\nits only asset is trust. MIT licensed.\n",
  "bytes": 5400,
  "sha": "5f029a84316179ba181cedd79dec04447d8c53f8978da059f985717cb38567dd",
  "repo_slug": "c47-inc/mcp-film",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_film_mcp_directory_c96a765b/readme"
}