Slice Timestamped YouTube & Shorts Transcripts into RAG Chunks
Test how hirepulse/youtube-transcript-scraper output (with segments[].start, duration, and text) transforms into overlapping vector-store documents with exact second-level YouTube deep-link citations (https://youtu.be/<id>?t=<seconds>) for LangChain, LlamaIndex, and Claude MCP agents.
1. Chunking Strategy & Window
Verified Sample Loaded: Real owner test run on
2026-10-15 (run_id: cGwKCKbXRTkojfKto, build 1.0.5) for video_id: "jNQXAC9IVRw" ("Me at the zoo", channel jawed, 19s, 39 words, 6 timestamped segments).
2. Actor JSON Output (Editable)
Valid JSON
3. Live RAG Document Chunks with Deep-Link Citations
4. Export Agent & RAG Pipeline Code