{"id":"janghyuckyun-mcp-youtube-intelligence","name":"mcp-youtube-intelligence","af_score":65.2,"security_score":35.0,"reliability_score":32.5,"what_it_does":"Provides an MCP server (stdio-based) and CLI to extract YouTube video metadata/transcripts, generate token-optimized summaries, perform topic segmentation and entity extraction, analyze comments with sentiment, and support channel monitoring, playlist analysis, search, and batch operations. Optional LLM providers (local via Ollama/vLLM/LM Studio, or cloud via OpenAI/Anthropic/Google) power the summarization/enrichment steps with a fallback to extractive summarization.","best_when":"You want structured YouTube intelligence with an agent-friendly MCP interface and you can run local or provide cloud LLM credentials, while benefiting from server-side transcript processing to reduce LLM token usage.","avoid_when":"You need well-documented operational guarantees (SLA, pagination semantics, strict rate-limit contracts) or you cannot supply/handle LLM credentials for the enrichment features.","last_evaluated":"2026-03-30T15:25:06.911347+00:00","has_mcp":true,"has_api":false,"auth_methods":["Optional cloud LLM API keys via environment variables (OPENAI_API_KEY, ANTHROPIC_API_KEY, GOOGLE_API_KEY)","Optional local LLM endpoints (Ollama/vLLM/LM Studio) configured via environment variables"],"has_free_tier":false,"known_gotchas":["LLM-based summarization may fail; documentation states an extractive fallback is used, but details of error recovery/retry are not specified.","YouTube URLs should be quoted in zsh due to '?' character handling (CLI gotcha).","Requires yt-dlp availability for transcript extraction; missing yt-dlp can cause failures.","For MCP via stdio, agents must launch the MCP server using the provided command (e.g., uvx) and pass required environment variables."],"error_quality":0.0}