| name | reveal |
| description | Reveal the solution for the active Mentor Mode objective after the learner explicitly asks. Use only when the user runs /reveal or explicitly asks to be shown the solution code. |
Reveal
Confirm Intent
Reveals are for genuine walls, not friction. Confirm once, briefly: this goes on the re-derive queue and comes back in a future warmup. If the learner has not yet made a real attempt and walked the hint ladder, offer the next hint level first - but an explicit "reveal it" wins. Never nag twice.
Unlock And Log
- In
mentor/guard.json: set reveal.active true, reveal.objectiveId to the active objective id, grantedAt to the current ISO timestamp.
- Append to revealQueue: objectiveId, a short concept label, revealedAt date, status "pending".
- Add or update the ledger row for the revealed concept now: status explained, confidence fail, interval at the first reviewIntervalsDays value, last reviewed today, next review tomorrow, with a note that it was revealed. The warmup re-derivation later upgrades it to re-derived.
Provide The Solution
Write or show the solution for the specific thing the learner is stuck on - the objective's scope, nothing extra. Explain it line by line at the learner's level: what each part does and why this shape rather than the obvious alternative. If writing into protected files, keep it minimal and idiomatic.
Re-lock
Immediately after delivering the solution: set reveal.active false and reveal.objectiveId null in mentor/guard.json. Reveals are one-shot. As a backstop, the hook ignores any reveal whose grantedAt is older than 2 hours.
After
The objective stays in-progress until the learner can explain the revealed code back. Do that explain-back now. The code returns in a future warmup for re-derivation from memory.
