Skip to main content
The Veedcrawl MCP server implements the Model Context Protocol so any compatible AI host — Claude, Cursor, Windsurf, and others — can call Veedcrawl’s APIs directly from a conversation or agent loop. Once installed, your agent can search for videos by topic, pull structured data from any public video on YouTube, TikTok, Instagram, X/Twitter, or Facebook, and explore creator profiles without needing to know a single video URL in advance.
The package is published as @veedcrawl/mcp on npm. You need a Veedcrawl API key to use it — get one at veedcrawl.com.

​
What the MCP server exposes

Six tools are registered automatically when the server starts. Each tool maps to a Veedcrawl API endpoint and returns a structured JSON result directly to your agent.
ToolWhat it returns
search_videosRanked list of public social videos matching a topic, hashtag, creator, or keyword — with URLs, stats, and thumbnails
get_video_metadataTitle, description, author, view/like/comment counts, tags, duration, thumbnails
get_video_transcriptFull spoken transcript with timestamped segments — uses native captions or AI-generated speech-to-text
extract_from_videoStructured answer to any prompt: hooks, CTAs, sentiment, topics, claims, quotes — guided by optional JSON Schema
get_tiktok_profileCreator bio, total video count, and up to 24 recent TikToks with full engagement stats per video
get_instagram_profileCreator bio, verified status, follower/following/post counts, and up to 24 recent posts (video, image, carousel) with per-post stats

​
Why search and profile tools change everything

Every other video API requires you to already know a video URL. That means your agent is stuck doing retrieval — fetching content you already found. search_videos unlocks topic-based discovery. Hand the agent a keyword or hashtag and it returns ranked public videos from any platform — no URLs, no prior knowledge required. Then it can chain directly into metadata, transcripts, and extraction. get_tiktok_profile and get_instagram_profile unlock creator-based discovery. Hand the agent a username and it comes back with who the creator is, how large their audience is, and exactly what they’ve been posting. For example, to find and analyze trending content from scratch: 
search_videos(q="skincare morning routine", platform="tiktok", limit=6)
→ 6 ranked TikToks with views, likes, and URLs
→ get_video_transcript(url=top_result.url)
→ extract_from_video(url=top_result.url, prompt="List every product mentioned and the claim made about each")
Or audit a creator from scratch: 
get_instagram_profile(username="hubermanlab", limit=24)
→ 283M followers, verified, last 24 posts with view/like counts
→ get_video_transcript(url=top_video.url)
→ extract_from_video(url=top_video.url, prompt="List every health claim made and the evidence cited")
Or run competitive analysis with no prep work: 
get_tiktok_profile(username=competitor_A, limit=24)
get_tiktok_profile(username=competitor_B, limit=24)
get_tiktok_profile(username=competitor_C, limit=24)
→ Compare posting frequency, avg views, top hashtags, engagement rates across all three

​
Next steps

Install the MCP server

Install @veedcrawl/mcp and configure it in Claude Desktop, Claude Code, Cursor, or Windsurf.

Tools reference

Full parameter and return value documentation for all five tools.