A skill that turns raw learning material (PDFs, recordings, images, video) into draft study notes and draft retrieval-bank questions for a guided-learning topic pack.
The agent prepares; the learner reviews and owns. This is conveyor work, not tutoring — the skill exists to remove the OCR/transcription/structure grind from studying, never to do the learning.
Clone the repo, then copy the inner study-prep/ folder into the tool's skills directory:
git clone https://github.com/jmstudyacc/study-prep-skill.git
cp -r study-prep-skill/study-prep <your-skills-dir>/| Tool | Skills directory |
|---|---|
| OpenCode | ~/.config/opencode/skills/ (user) or .opencode/skills/ in a repo |
| Claude Code | ~/.claude/skills/ (personal) or .claude/skills/ in a repo |
| OpenWork | .agents/skills/ in the workspace (as used in this vault) |
Start a new session to pick it up. The path must be <skills-dir>/study-prep/SKILL.md.
- Extracts per material type: PDF text/OCR, audio transcription, image OCR, video audio + keyframes.
- Produces draft notes in
notes/_prep/— one file per concept, sources cited, markedprepared_by: agent/status: draft. - Produces a draft question bank from the material — never merged into the real bank until the learner reviews it.
- Hands off with a short "what to review" summary. No teaching.
The contract in one line: drafts are marked and staged; nothing counts until the learner adopts it; the agent never does the learning. Full contract and anti-patterns are in SKILL.md.
Designed as the conveyor companion to guided-learning:
- Invoked from guided-learning sessions when source material arrives with a topic.
- Output targets the pack's
notes/_prep/staging area; adopted notes/questions flow into the pack perpack-formats.md. - Reuses the vault's media-pipeline OCR/STT tooling when reachable; never depends on it.
MIT — see LICENSE.