MCP Server for AI Assistants
The Model Context Protocol (MCP) is the open standard that lets AI assistants use external tools. The Greenlight MCP server connects Claude, ChatGPT, Cursor, and any other MCP client directly to all three Greenlight catalogues — film development packages, published books, and audiobooks — through one remote endpoint.
https://www.greenlight.film/api/mcp
How it works
Point your assistant at the server
The server is hosted — there is nothing to install or run. Add the endpoint URL to any MCP-capable host (Claude, ChatGPT, Cursor, VS Code, or your own agent).
Sign in once
OAuth hosts discover the server automatically and walk you through a Greenlight sign-in (OAuth 2.1 with PKCE). Prefer keys? Generate a gl_pk_… key and send it as a Bearer token instead.
Your assistant calls tools
The host lists 22 tools and calls them as you chat — browse projects, pull screenplays, read book chapters, stream audiobook chapters, browse castable characters, and (for casting partners) submit performer profiles. Your access tier follows your plan.
Under the hood: MCP Streamable HTTP with stateless JSON. Requests are metered against the same hourly rate limits as the REST API — Free 10 req/hr, Starter 100, Growth 1,000, Enterprise 10,000. Your tier follows your subscription plan.
Connect from your client
One command in your terminal; Claude Code walks you through the OAuth sign-in on first use:
claude mcp add --transport http greenlight https://www.greenlight.film/api/mcp
Settings → Connectors → Add custom connector → paste the server URL. Sign in with your Greenlight account when prompted and click Approve.
https://www.greenlight.film/api/mcp
Settings → Connectors → enable Developer mode → Create → paste the server URL with authentication set to OAuth. Sign in when prompted. Works with deep research — see the note below.
https://www.greenlight.film/api/mcp
Add an HTTP MCP server pointing at the endpoint. In Cursor, add this to .cursor/mcp.json:
{
"mcpServers": {
"greenlight": {
"url": "https://www.greenlight.film/api/mcp",
"headers": { "Authorization": "Bearer gl_pk_your_key_here" }
}
}
}OAuth (Claude, ChatGPT)
Hosts discover the server via /.well-known/oauth-protected-resource and run OAuth 2.1 with PKCE. Sign in with your Greenlight account and click Approve — no key handling, and your access tier follows your account's plan.
API key (Cursor, SDKs)
Generate a gl_pk_... key from your API keys page and send it as a Bearer token in the client's MCP config, or as authorization_token in the Claude API / OpenAI Responses API MCP config.
ChatGPT deep research
The search and fetch tools follow the OpenAI connector contract: search returns ids like project:204, book:{slug}, and audiobook:{slug}, and fetch resolves them to full documents. Results cite canonical URLs on greenlight.film, greenlight-publishing.com, and greenlight-audio.com, so deep-research reports link straight back to the catalogue.
The 22 tools
| get_license_status | Tier, rate limit usage, catalogue versions (unmetered) |
| list_projects | Browse the film catalogue with filters and sorting |
| search_projects | Search projects by title keyword |
| get_project | Full metadata for a single project |
| get_reviews | Critic reviews for a project |
| get_assets | All image/video URLs for a project |
| get_deliverable | One deliverable's full text (screenplay, treatment, ...) |
| get_project_package | Complete project with all deliverables |
| list_books | Browse the published novels |
| get_book | Book detail, author bio, chapter list, store links |
| get_book_chapter | Chapter text — free-preview chapters at any tier |
| export_book | Full book with all chapter content (billed as 5 requests) |
| list_audiobooks | Browse audiobooks by language and genre |
| get_audiobook | Audiobook detail with chapter durations |
| get_audiobook_chapter_audio | Streamable audio URL for one chapter |
| list_characters | Castable roles parsed from a project's Character Bible — name, age, role, design summary, voice, arc |
| search_actors | Browse/search the casting registry: human and synthetic performer profiles |
| get_actor | One performer profile by slug, incl. synthetic-performer provenance |
| submit_actor | Casting partners add/update performer profiles (write; attributed to your key) |
| get_catalog_stats | Aggregate statistics across projects and book genres |
| search | Unified search across all three catalogues (ChatGPT deep-research contract) |
| fetch | Fetch a search result by id — project:241, book:slug, audiobook:slug |
All catalogue tools are read-only; submit_actor is the one write (casting partners, growth tier, attributed to your key). Search covers titles (projects) and title/description/genre (books, audiobooks) — not full text. A local stdio package (@greenlight/mcp-server) remains available for legacy setups.
Prefer raw HTTP? The same catalogue is available over the REST API with 34 endpoints.