Overview
As the leading provider of advanced entertainment experiences across linear, streaming and FAST, Gracenote has served the industry as the source of truth across the entire video ecosystem. We are expanding that expertise into the AI domain by providing tools that allow LLMs to validate and enrich their responses with human-verified, up-to-date, and accurate data.
Gracenote's Video Model Context Protocol (MCP) Server ensures that LLMs are grounded with real-world information to reduce hallucinations and ensure that LLMs have access to Gracenote entertainment metadata to make the best possible responses to LLM prompts. By connecting LLMs to Gracenote's knowledge base, responses from LLMs for each real-time query are verified, corrected, and enriched to provide responses that are based on the leading source of truth data spanning the entire video ecosystem.
The Gracenote Video MCP Server works with Large Language Models (LLMs) from providers like Anthropic (Claude), OpenAI (GPT), and Google (Gemini). The LLM interprets user queries and calls MCP tools to retrieve Gracenote metadata.
This guide provides a comprehensive overview of the Gracenote Video MCP system. It details the architecture, workflow, and core concepts necessary for developers to successfully integrate and customize their LLM Agent using Gracenote's proprietary metadata.
About this Release
The General Availability (GA) release expands the MCP Server with global data support and new tools for comprehensive video metadata access.
Features
This GA release provides production-ready capabilities for video metadata grounding and enrichment:
- Entity Resolution: Disambiguate movies, TV series, and episodes with global data support
- Streaming Availability: Real-time "where to watch" information for streaming services
- Metadata Tools: Access comprehensive title information, images, and alternate language versions
- Usage Monitoring: Customer dashboard for tracking usage transparency
Gracenote Sports, Music, and Studio System data sets will be made available via dedicated MCP interfaces in future releases.
Response Data
Responses from the MCP server includes Gracenote video core metadata:
- Episodes: individual programs (TV episodes, one-off shows, etc.)
- Series: serialized programs (episodic TV shows, miniseries, talk shows)
- Movies: theatrical-release and made-for-TV movies
Response data includes program data such as descriptions, genre, cast, crew, awards, image links, TMSIds, among many others.
Not Included
The following features and data are not included:
- Linear TV schedules, lineups, and station data
- Restricted/Approval-only catalogs
- FAST channel data that is not already included in Linear TV schedules
- Additional descriptors: Watch Prompts, Video Descriptors, Popularity scores
- Filter on specific catalogs
- Historical availability data or "coming soon" titles
- Sports data (available via separate Sports MCP Server)
- Person-specific search tools
- Content Types: Returns only movies (Feature Film, TV Movie), TV series, and TV episodes. Sports events, previews, news, specials, and bonus content are excluded.
- Availability Scope: Streaming availability returns "today" only (current availability at time of query).
- Image URLs: Provided for demo purposes only. Production implementations must cache images locally.
Next Steps
- Gracenote MCP Demo - Try the live demo
- Quick Start (CLI/Proxy) - Test locally with Claude Desktop or MCP Inspector using the STDIO proxy
- Connecting to MCP - Build your own integration with direct HTTPS authentication
- Use Cases - See what you can build