Study material

yat:media prepares sources for a learning env. It is setup work, not a learning session, so it may run commands freely.

What you say

You say	What happens
extract this pdf	Text out, with the page count and any pages that came back empty
absorb this book	Extract, index by page or section, and register it in the graph
convert this video / convert this audio	A native ffmpeg conversion, reporting the exact command

Where things go

Drop sources in media/<env>/. Outputs land in media/out/, named after the source, and are never overwritten silently. Absorbing a text source also writes media/out/<stem>.index.md, which is what yat:read cites by page.

Pages that come back empty, usually scanned images, are rendered and flagged rather than passed off as complete.

Files it touches

Reads	Writes
media/<env>/	media/out/, media/out/<stem>.index.md
	memory/ontology/graph.jsonl (a resource and its chapters)

Terminal: allowed. Needs pdftotext and ffmpeg.

Source

harness/skills/media/SKILL.md