Developers
API and MCP docs
Flatcast has an MCP server for AI assistants and a REST API for your own code. Both read the transcripts, summaries and comments Flatcast already made, so they use none of your AI hours. Both are on every paid plan.
Connect Claude or ChatGPT
The MCP server is at https://flatcast.ai/mcp. Add it as a custom connector and sign in to Flatcast. You choose which workspaces it can read, and whether it may also post comments and edit videos.
- Claude: Settings → Connectors → Add custom connector, paste the address, then Connect.
- ChatGPT: Settings → Apps & Connectors → Create, paste the address and choose OAuth.
- Claude Code: make an API key in workspace Settings, then run:
claude mcp add --transport http flatcast https://flatcast.ai/mcp \
--header "Authorization: Bearer fc_live_…"Cursor and other MCP clients take the same address and header in their MCP settings.
Tools
search_videos | Search what was said, and titles, summaries and chapters |
list_recent_videos | Newest videos, with filters by date and owner |
get_video | Summary, chapters with links, counts and share link |
get_transcript | Timestamped transcript, in pages |
list_comments | Comments and reaction counts |
get_viewers | Who watched (owner or admin) |
list_workspaces | Workspaces the connection can read |
add_comment | Post a comment or reply as you |
update_video | Change title, summary or chapters (owner or admin) |
set_video_privacy | Make a share link public or workspace-only (owner or admin) |
search, fetch | The two tools ChatGPT's deep research uses |
REST API
Base URL https://flatcast.ai/api/v1. Send a personal API key from workspace Settings (or an OAuth access token) as a Bearer token. Requests and responses are JSON.
curl https://flatcast.ai/api/v1/search?q=billing+bug \
-H "Authorization: Bearer fc_live_…"curl -X POST https://flatcast.ai/api/v1/videos/vid_123/comments \
-H "Authorization: Bearer fc_live_…" \
-H "Content-Type: application/json" \
-d '{"body": "Looks good, ship it", "atSeconds": 42}'Endpoints
| Request | What it does | Needs |
|---|---|---|
GET /workspaces | Workspaces the key can reach | videos:read |
GET /videos | Newest videos. Query: workspace, since, until, owner=me, limit (max 50) | videos:read |
GET /search?q=… | Search transcripts, titles, summaries and chapters. Same filters, limit max 20 | videos:read |
GET /videos/{id} | Details, full summary, chapters, counts and share link | videos:read |
PATCH /videos/{id} | Change title, summary, chapters or privacy (public or workspace). Owner or admin | videos:write |
GET /videos/{id}/transcript | Timestamped transcript. Query: start, end (seconds), cursor | videos:read |
GET /videos/{id}/comments | Comments, replies and reaction counts | comments:read |
POST /videos/{id}/comments | Comment as the key's owner: body, atSeconds, parentId | comments:write |
GET /videos/{id}/viewers | Who watched and how much. Owner or admin | comments:read |
A video {id} is its vid_… id or the share id at the end of its link. Times are seconds in the version viewers watch, after any edits. Dates are ISO 8601, like 2026-11-20.
Keys, permissions and limits
- A key belongs to one person and one workspace, and sees only what that person can see. It stops working if they leave the workspace. Admins can revoke any key in Settings.
- Keys are read only or read and write. Writing means posting comments as the key's owner, and changing videos they own (or any video, for admins).
- Limits: 60 requests a minute per key, 20 searches a minute, and 5,000 requests a day per workspace. Over a limit you get
429; wait and try again. - Errors look like
{"error": {"code": "video_not_found", "message": "Video not found"}}with the matching HTTP status. A video you can't see is always a404.
Questions or something missing? Email us.