YouTube transcript aggregation, summarization, and digest system with MCP server for AI agents.
View on GitHub ↗Glean transcripts from YouTube channels, store them in SQLite, summarize via LLM, and expose the data as an MCP server.
- Track YouTube channels and automatically fetch new video transcripts - Dual transcript providers: InnerTube API (fast, pure Go) with yt-dlp fallback - Automatic channel ID and name resolution from handles/URLs - Language fallback: tries all configured transcript languages before failing - Retry with exponential backoff on transient failures - Rate limiting protection with configurable fetch delay - Summarize transcripts via any OpenAI-compatible API (OpenAI, Ollama, LM Studio, etc.) - Full-text search across stored transcripts - MCP server for AI agent integration (Claude Desktop, Cursor, etc.) - Configurable data retention and pruning
- yt-dlp (optional, for transcript fallback): pip install yt-dlp - LLM API (for summarization): OpenAI API key, or a local Ollama/LM Studio instance
On first run, YTGlean auto-generates a config file at ~/.config/ytglean/config.yaml. Edit it to set your API key and preferred settings.
On first run, YTGlean creates ~/.config/ytglean/config.yaml with sensible defaults. Edit this file to customize settings.
Config is loaded from (in order of priority):
1. --config flag 2. ~/.config/ytglean/config.yaml 3. ./config.yaml
Channel inputs accept: - Handles: @Fireship - Full URLs: https://www.youtube.com/@Fireship - Channel IDs: UCsBjURrPoezykLs9EqgamOA - Names (for remove): Fireship
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/Crymfox--YTGlean.html)
Your agents already solved this. deja finds it — retroactive search and MCP recall over the session logs your coding agents already wrote. Cross-agent, SSH-portable, one zero-dep binary, fully local.
enowdev/succubusCross-agent coordination for AI coding agents. One daemon, one database — so multiple agents in one repo can see each other's plan, tasks, and file claims.
hoangann2000/figma-mcp-consoleMCP server that lets your AI assistant read and write Figma in real time — no API token, no rate limits. Works with Claude, Cursor, Copilot, Codex and any MCP client.
BeppeTemp/cartographerMCP governance server in Go for the Agentic Wiki — knowledge that composes, not that you query
Evermemoai/evermemoA tiny, universal memory engine for humans and AI agents — CLI, HTTP API, MCP server, and shared memory hub in one small Go binary
agentcathq/agentcat-go-sdkAgentCat is an analytics platform for MCP server owners 🐱
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.