Video Intelligence API: Integrate AI Video Analysis into Your Stack
VidForgeX exposes its full analysis capability as an API — REST for traditional integrations and MCP (Model Context Protocol) for AI agent workflows. Submit video, retrieve structured results, and query your video library programmatically.
API architecture
VidForgeX uses a two-surface API architecture:
Traditional HTTP endpoints via the VidForgeX backend. Base URL: api.vidforgex.com. Action-based routing with Bearer token auth.
11 AI-native tools at mcp.vidforgex.com. Compatible with any MCP client (Claude, Cursor, LangChain MCP adapter, custom agents). Same vfx_ API key auth.
The 11 MCP tools
| Tool | Description |
|---|---|
analyze_session | Submit video URL or ID for AI analysis |
get_session_result | Retrieve structured analysis result JSON |
ask_about_session | AI chat scoped to a single video session |
ask_about_person | AI chat drawing from all sessions tagged to a person |
compare_profiles | Side-by-side comparison of 2–10 persons |
get_person_progression | Timeline of metrics per person (1–50 sessions) |
find_person_in_video | Find timestamps where a specific person appears |
get_team_summary | Aggregated metrics over a configurable period (1–365 days) |
get_sop_compliance | SOP compliance results (Professional + TMA vertical) |
tag_person_to_session | Tag or create a person Tracking Profile |
generate_sop_document | Generate SOP from rule template (Professional + TMA vertical) |
Authentication and key management
API keys are created in VidForgeX under Settings → API Keys. Key format:vfx_ followed by 40 hex characters (160 bits of entropy). Only the prefix and a cryptographic hash are stored server-side — the full key is shown once on creation. Keys support:
- Custom name and expiry (never, 30, 90, or 365 days)
- Scope restriction (analysis, chat)
- Org scope (single org, selected orgs, or all user orgs)
- Instant revocation via the UI
Rate limiting
Rate limits are enforced globally at the proxy layer using a sliding window algorithm. This ensures limits are consistent across all instances — horizontal scaling cannot be used to bypass limits.
When a rate limit is exceeded, the API returns a JSON-RPC error -32004 with aretry_after_ms field indicating when the next request can be made.