The VideoDB Director MCP is a Model Context Protocol server that integrates with the Director backend to provide a single tool for multiple AI-driven workflows. It is designed to run in isolated environments via uvx, enabling development and testing without impacting production. You start it with uvx videodb-director-mcp --api-key=VIDEODB_API_KEY, and keep installations current by clearing the UV cache and updating to the latest MCP package with uvx videodb-director-mcp@latest --api-key=<VIDEODB_API_KEY>. This MCP server works alongside the VideoDB Agent Toolkit, exposing standardized context to LLMs and agents and supporting the llms-full.txt (comprehensive context) and llms.txt (lightweight discovery) files. Automated maintenance is orchestrated by GitHub Actions, updating SDK, docs, and notebook examples, and synthesizing a master llms-full.txt along with the discovery llms.txt. A central config.yaml governs customization, including which files contribute to llms-full.txt and how they are assembled. Together, these components ensure AI applications always operate with accurate, up-to-date context across workflows.