We use essential cookies for authentication and optional analytics cookies to improve the platform. Privacy Policy

    MCP Video API

    VidForgeX MCP: Video Intelligence for AI Agents

    The VidForgeX MCP server gives Claude, Cursor, and any Model Context Protocol client direct access to 11 video intelligence tools — analyze footage, search your library, track persons, and generate reports without leaving your LLM workflow.

    What MCP gives you that REST does not

    A REST API requires you to write code that calls it. The MCP server lets the LLM call it directly. When Claude is analyzing a coaching session and needs to compare two players, it does not need a human to write an API call — it calls compare_profiles automatically, with the correct parameters, in the context of the conversation.

    This enables entirely new workflows: a coach asks Claude to "prepare a scouting report on our next opponent based on the last 3 videos I uploaded" — Claude calls get_session_resultthree times, synthesizes the analysis data, and produces the report without the coach touching any API.

    MCP configuration

    Claude Desktop config (claude_desktop_config.json):

    Add VidForgeX under the mcpServers key with URL https://mcp.vidforgex.com, transport sse, and your vfx_ API key as the Authorization Bearer token.

    Tool reference

    ToolKey parametersReturns
    analyze_sessionvideo_id, analysis_type, organization_idJob ID for polling
    get_session_resultvideo_id, analysis_typeFull structured analysis JSON
    ask_about_sessionvideo_id, question, history[]SSE or JSON text answer
    ask_about_personperson_id, questionAnswer grounded in person's video history
    compare_profilesperson_ids[] (2–10), metric_focusSide-by-side comparison table
    get_person_progressionperson_id, limit (1–50)Metric timeline across sessions
    find_person_in_videovideo_id, reference_image_urlSighting timeline with timestamps
    get_team_summaryorganization_id, period_days (1–365)Aggregated team metrics
    get_sop_compliancevideo_id, rule_template_idCompliance score + violations
    tag_person_to_sessionvideo_id, person_name or person_idTag confirmation or disambiguation list
    generate_sop_documentvideo_id, template_idFormatted SOP document

    Rate limiting

    Rate limiting for the VidForgeX API is enforced globally at the proxy layer using a distributed counter, ensuring limits are consistent across all instances simultaneously — horizontal scaling cannot be exploited to exceed limits.

    Tiers: standard=60/min, elevated=300/min, unlimited=no limit. When exceeded: JSON-RPC error -32004 with retry_after_ms.

    MCP video APIModel Context ProtocolClaude MCPVidForgeX MCPAI agent video toolsAI video integrationLLM video toolsAnthropic MCPCursor MCPvideo AI agent

    Frequently Asked Questions

    What is the Model Context Protocol (MCP)?
    MCP is an open standard developed by Anthropic that defines how AI models (like Claude) communicate with external tools and data sources. An MCP server exposes a set of typed tools that LLM clients can discover and call. VidForgeX implements an MCP server at mcp.vidforgex.com, making its video intelligence capabilities available to any MCP-compatible AI client.
    How do I connect Claude to VidForgeX?
    In your Claude MCP configuration, add an entry for VidForgeX: server URL mcp.vidforgex.com, transport SSE or HTTP, authentication Bearer with your vfx_ API key. Once configured, Claude can call all 11 VidForgeX tools in any conversation where video intelligence is needed.
    What can an AI agent do with VidForgeX MCP tools?
    An AI agent with VidForgeX MCP access can: submit videos for analysis, wait for completion and retrieve results, search a video library by person or concept, compare athletes or workers across sessions, generate SOP documents from observed procedures, query compliance data, and produce trend analysis by calling get_person_progression.
    What authentication does the MCP server use?
    The VidForgeX MCP server uses vfx_ API key authentication. Bearer tokens are validated on every request. Rate limiting is enforced globally at the proxy layer before requests reach the function, ensuring limits apply across all instances simultaneously.