TranscriptAPI

TranscriptAPI is a hosted REST API and Model Context Protocol (MCP) server that gives developers and AI agents reliable access to YouTube data — transcripts, video search, channel browsing, and playlist extraction — without requiring a Google YouTube Data API key or dealing with YouTube's IP blocks. The service is operated by ZeroPointRepo under the TranscriptAPI brand and is actively used in production, with the homepage stating it processes 15M+ transcripts per month and 500K+ transcripts daily at a median response time of 49ms.

What It Is

TranscriptAPI sits in the category of unified YouTube data APIs designed for AI pipelines and automation workflows. Its core job is to reliably fetch YouTube transcripts and metadata at scale, bypassing the infrastructure challenges (YouTube blocks major cloud IPs, yt-dlp breaks frequently) that make direct scraping impractical. Every feature — transcripts, search, channel video lists, playlist extraction, and new-upload monitoring — is accessible via a single REST API, an MCP server compatible with Claude, ChatGPT, Cursor, and VS Code, or installable "agent skills" for runtimes like OpenClaw, Hermes Agent, Claude Code, and Cursor.

Three Access Paths

TranscriptAPI offers three distinct integration modes, each targeting a different audience:

• REST API: Standard HTTP endpoints authenticated with a bearer token. Integrates with n8n, Make, Zapier, or custom code. Each successful request costs one credit; channel handle resolution and latest-upload checks are free.
• YouTube MCP: A Model Context Protocol server at https://transcriptapi.com/mcp that plugs directly into Claude, ChatGPT, Cursor, VS Code, and 20+ other MCP-compatible clients. One-line install via Claude Code or a JSON config block.
• Agent Skills: An open-source MIT-licensed skill package (ZeroPointRepo/youtube-skills on GitHub) installable via npx skills add or npx clawhub@latest install youtube-full. On first run, the agent automatically registers a free API key using the user's email and an OTP — no manual dashboard visit required.

Endpoint Coverage

The API exposes six functional areas:

• Get Transcript: Returns full transcript text with per-segment timestamps and video metadata (title, duration, language).
• Search YouTube: Searches for videos, channels, or playlists with filters for type and result count.
• Search Inside a Channel: Scopes a keyword search to a specific channel by @handle, URL, or UC ID.
• Browse Channel Videos: Paginates through a channel's full upload history (100 videos per page).
• Track New Uploads: Free RSS-based endpoint that returns the latest videos from any channel without consuming credits.
• Extract Playlist Videos: Returns all videos in a playlist, supporting PL, UU, LL, FL, and OL prefix formats.

Workflow Library

The TranscriptAPI team publishes a companion open-source workflow library (ZeroPointRepo/youtube-workflows on GitHub, MIT-licensed) with 58 ready-to-use automations covering content repurposing, RAG ingestion, competitive intelligence, education, and data pipelines. Workflows target AI agents, n8n, Make, and Zapier. Examples include single-video-to-SEO-blog-post, channel-to-Pinecone-RAG-ingestion, earnings-call tracker with Slack digest, and playlist-to-quiz/flashcard generator.

Agent Skill Ecosystem

The youtube-skills repository on GitHub (957 stars, 106 forks as of the last recorded update) ships 12 installable skills. The recommended youtube-full skill bundles transcripts, search, channels, and playlists. Narrower skills (transcript, youtube-search, youtube-channels, youtube-playlist) are available for agents where smaller context improves performance. The repository is listed in the awesome-hermes-skills directory and is compatible with OpenClaw, Hermes Agent, Claude Code, Cursor, Antigravity, Cline, Codex, Goose, AMP, Kiro, Roo, OpenCode, Gemini, and GitHub Copilot.