blisolver-video-ingestion

Operate the BliSolver ingestion pipeline for bilibili.com and YouTube videos - diagnose the runtime, probe metadata cheaply, run an ingest, and inspect or validate the resulting Atlas bundle. Use when an Agent needs video acquisition, caption-versus-Whisper decisions, frame/vision/OCR context, danmaku or interaction provenance, provider troubleshooting, or schema-1.1 bundle handling. Do not use it for downstream summarization or entity extraction.

License

MIT. LICENSE.txt has complete terms.

Compatibility

Requires Python 3.11+ and a Python environment holding BliSolver's dependencies. Media stages additionally require ffmpeg, whisper.cpp with a GGML model, a JavaScript runtime for YouTube, LM Studio for vision, and an isolated OCR environment.