How to extract YouTube transcript in Claude Desktop
YouTube transcripts are buried behind menus, split into tiny timestamped chunks, and offer no export. Claude Desktop can extract them through three Firecrawl integrations:
| Method | Install | Output | Best for |
|---|---|---|---|
| MCP server | Add to MCP config, restart | Inline in chat | Most users — works with all 13+ Firecrawl tools |
| Claude plugin | claude plugin install firecrawl@claude-plugins-official | Inline in chat | Fastest setup, no config files |
| CLI | npx -y firecrawl-cli@latest init --all --browser | File on disk | Long transcripts — keeps token usage low |
MCP server — add npx -y firecrawl-mcp to your Claude Desktop MCP config with your API key, restart, then paste a YouTube URL and ask Claude to extract the transcript.
Plugin — after installing, use /firecrawl:scrape <URL> or just ask in plain English. The /firecrawl:agent command can autonomously navigate and extract transcripts from complex pages.
CLI — run firecrawl scrape <URL> to write the transcript to disk. Claude Desktop can read the file without loading the full text into the context window, which matters for hour-long videos.
All three handle JavaScript rendering and page structure changes automatically. Get a free API key at firecrawl.dev.