Let your app — or your AI agent — download video.
One link in, files out. Point agents at the MCP endpoint, or call REST directly. Machine spec: openapi.json · agent card: llms.txt.
Option A — MCP (recommended for agents)
A stateless Model Context Protocol endpoint. Add it to Claude Desktop, Cursor, Windsurf or any MCP client:
{
"mcpServers": {
"gtv": { "url": "https://getthevid.com/api/mcp" }
}
}Then the agent gets four tools — fetch_video_info, download_video, merge_video, extract_audio — each taking a page url (plus optional carousel slide). So a user can literally say “download this for me: <link>” and the agent handles the rest. Point them at https://getthevid.com (agents: use the host of the page you're reading).
Option B — REST
| Call | Does |
|---|---|
GET /api/fetch?url=…&slide=… | Variants best-first, images, author, stats, slide info |
GET /api/download?url=… | Attachment bytes for a resolved media URL |
GET /api/merge?url=… | MP4 with sound (15–120s, long timeout) |
GET /api/audio?url=… | MP3 bytes |
GET /api/health | Service + yt-dlp/ffmpeg status |
# quick test (replace the video URL)
curl "https://getthevid.com/api/fetch?url=https://x.com/spacex/status/1845922312207712396" | head -c 400Rules of the road
Public only
Private, deleted and login-walled posts return 404. Only fetch media the user owns or may save.
Respect limits
Fetch 30/min, download 40/min, merge 6/hr, MP3 10/hr, MCP 60/min per IP. Back off on 429.
Attribute fairly
Keep filenames/credits intact and respect platform Terms and copyright.
API FAQ
MCP or REST — which should my agent use?
MCP if the client supports it (typed tools, less parsing). REST otherwise — same engine underneath.
Do I need an API key?
No. Fair-use rate limits apply; heavy users should self-host (zero dependencies).
How do carousel slides work?
Pass slide=N (1-based video number). Responses echo slide and slideGroups.