ingest-video

What problem does it solve?

Local video files such as meetings, presentations, and screen recordings are often locked as opaque media that are hard to search, summarize, or integrate into a knowledge base. This Skill automates transcription, speaker attribution, and screenshot extraction so video content becomes searchable, citable, and summarized inside your wiki.

Core Features & Use Cases

• Transcribe audio from local videos using a tiered provider approach (local Whisper, OpenAI Whisper API, or AssemblyAI) with diarization where available.
• Extract scene-change keyframes and fallback to I-frame extraction, cap keyframes, and save screenshots into the vault assets.
• Produce a raw transcript file with frontmatter metadata and create a structured wiki source note that includes an overview, attendees, key decisions, action items, key takeaways, embedded screenshots, and a collapsible full transcript.
• Provide a post-extraction report detailing provider used, keyframe count, diarization availability, and a short content summary.
• Use Case: Convert recorded meetings into actionable wiki pages with tasks and decisions, or turn lecture videos into searchable study notes.

Quick Start

Use the ingest-video skill to transcribe the local file meeting.mp4 and create a wiki source note with keyframes and the full transcript.