wizzy-mcp-tmdb

MCP TMDB Server (JavaScript) Overview - This repository includes an MCP (Model Context Protocol) server that allows an AI client to search TMDB (The Movie Database) and fetch details for items. - Implemented in JavaScript (no TypeScript), file: mcp-tmdb-server.js - Tools exposed: - search_tmdb: Multi-search across movies, TV, and people - search_tmdb_movies: Targeted search for movies with optional year filter - get_tmdb_details: Fetch details for movie/tv/person by id Prerequisites - Node.js 18+ (for global fetch) - TMDB_AUTH_TOKEN: a Bearer token used with the TNL TMDB proxy (production-api.tnl.one). Ask your admin for the token, then set it as environment variable TMDB_AUTH_TOKEN. Install 1) Install dependencies: npm install Start the MCP server - Windows PowerShell (current environment): $env:TMDB_AUTH_TOKEN="YOUR_TNL_PROXY_BEARER_TOKEN"; npm start - macOS/Linux: TMDB_AUTH_TOKEN="YOUR_TNL_PROXY_BEARER_TOKEN" npm start How to use with an MCP client - This server communicates over stdio. Configure your MCP-compatible client (e.g., Model Context Protocol capable IDE or chat client) to start the command: Command: node mcp-tmdb-server.js Env: TMDB_AUTH_TOKEN=YOUR_TNL_PROXY_BEARER_TOKEN Quick local test (discover/tv) - After starting the server, you can manually test the proxy token is working with curl (bypassing MCP) using the exact endpoint the tools would call: Windows PowerShell: curl -H "Authorization: $env:TMDB_AUTH_TOKEN" -H "Accept: application/json" ` "https://production-api.tnl.one/service/tmdb/3/discover/tv?language=en&page=1&with_watch_providers=8&sort_by=release_date.desc&watch_region=IS" macOS/Linux: curl -H "Authorization: $TMDB_AUTH_TOKEN" -H "Accept: application/json" \ "https://production-api.tnl.one/service/tmdb/3/discover/tv?language=en&page=1&with_watch_providers=8&sort_by=release_date.desc&watch_region=IS" - Expected: a JSON payload listing TV results. If you get 401/403, verify TMDB_AUTH_TOKEN is set and valid. Tool schemas and examples - search_tmdb Input JSON: { "query": "dune", "page": 1, "language": "en-US", "include_adult": false } Returns JSON with compact results: id, media_type, title, date, overview, etc. - search_tmdb_movies Input JSON: { "query": "mission impossible", "year": 1996 } - get_tmdb_details Input JSON: { "type": "movie", "id": 438631, "append": "credits,images" } Junie usage guidelines - Purpose: Provide the AI with commands to search on TMDB via MCP tools. Prefer multi-search (search_tmdb) when you don't know the media type. - Disambiguation: When the AI receives multiple results, it should pick by id and call get_tmdb_details for more info. - Parameters: - language: Prefer en-US unless the user requests another locale. - Pagination: If results seem truncated, increment page. - Adult content: Set include_adult=false unless the user explicitly requests adult content. - Error handling: If the server returns an error about TMDB_AUTH_TOKEN, ensure the environment variable is set before retrying. Notes - This server is intentionally minimal to satisfy the requirements. - Extend with more tools if needed (e.g., trending, discover endpoints).