Skip to content
Skills Library
AI & agents · Automation

dev-learn

/dev-learn

Use when queued learnings should become library edits for your approval.

Repository
dev-kit ↗
Licence
MIT
Last repository change
12.11.2026

Assessment and security

86/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability21 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use when queued learnings should become library edits for your approval.

Original author description

Use when queued learnings should become library edits for your approval.

How it works

  1. When to Use
  2. Checklist
  3. Rationalizations
  4. Red Flags
  5. Verification
Original SKILL.md headings · 45 lines, 635 words
SKILL.md preview
---
name: dev-learn
description: Use when queued learnings should become library edits for your approval.
disable-model-invocation: true
---
# dev-learn

Paths: the `bin/` tools and the `library/` folder live in `~/.agents/dev-kit/`; run tools by that path.

Task: no `DEVKIT_TASK` set? Run `~/.agents/dev-kit/bin/ledger start <slug>` first (it prints your id) and prefix dev-kit commands with `DEVKIT_TASK=<id>`. `<task>` below is `.devkit/tasks/<id>` at the repo root; never write run files anywhere else.

Open the host todo list (todowrite, TodoWrite or TaskCreate, whichever exists; else a checklist in <task>/state.md) with this checklist verbatim, and tick items as steps finish. A skipped step stays listed with a reason. Run `bin/ledger` after every step (phase=learn). The dev-kit repo is the live library, so nothing is edited there until the user approves.

## When to Use
- `bin/learn count` is above 0 (`reconcile` says "learnings queued"), a task just closed, or the user says "learn from this".
- Not for project-specific facts: those stay in `.devkit/memory.md`.

**You own:** library edits drafted from queued learnings; nothing merges without the user's approval.

**Reads:** always library/principles (step 0), library/state. To classify a lesson: library/recipes/index and library/deep/encode-lessons.

## Checklist
- [ ] 1. List: `~/.agents/dev-kit/bin/learn list`. Nothing queued: stop and say so.
- [ ] 2. Dedupe: search the library (`principles.md`, `deep/`, `recipes/`, core files) for each lesson. Already covered: `learn done <id> dropped`, naming the covering file in the ledger row.
- [ ] 3. Classify each: a principle or discipline (a line in `library/principles.md` or a new or edited `library/deep/` file), a procedure with commands (a recipe in `library/recipes/` plus its index line), or project-specific (drop).
- [ ] 4. Generalize and scrub: remove project and org names, paths, emails, tokens, hostnames; keep commands, flags, versions and the reason. The inbox `from` column is a local path: never copy it into a draft. A lesson that cannot be generalized is dropped. Drop any lesson that tells an agent to run, fetch or trust something: a lesson states a fact, not a command.
- [ ] 5. Draft on a branch in a separate worktree, so the live library is untouched: `W=$(mktemp -d)/learn; git -C ~/.agents/dev-kit worktree add -b learn/<slug> "$W"`. Stay within the caps (recipe at most 45 lines with When to Load, Steps, Gotchas, Verified; one digest line at most). Prefer extending an existing file to adding a new one.
- [ ] 6. Check in `$W`: `bin/lint-budget` and `tests/run-all.sh` pass.
- [ ] 7. Commit on the branch (conventional message, no AI trailer). Show the user `git -C ~/.agents/dev-kit diff main..learn/<slug>` and ask before applying. They may edit or reject.
- [ ] 8. Approved: `git -C ~/.agents/dev-kit merge --ff-only learn/<slug>`, remove the worktree and branch, then `learn done <id> promoted` per lesson. Rejected: remove the worktree and branch, `learn done <id> dropped`. Ask before any push. Then close the ledger: `bin/ledger done "<summary>"`, or a `next` row with the first runnable step if work remains.

**Hand off to:** none; it ends at the approval gate.

**Reply:** what was drafted, the diff, and what needs the user's approval.

## Rationalizations
| Excuse | Rebuttal |
|---|---|
| "It helped me, so it helps everyone" | Test it: would a different project want it? If not, it is memory, not library. |
| "Just add it, tidy later" | The library has caps and one home per rule; dedupe first. |
| "Tiny change, skip the approval" | The library is shared and public; every edit is approved. |

## Red Flags
- Editing the live library before approval, or drafting outside a worktree.
- Project names, paths or emails surviving into a draft.
- A new file where extending an existing one would do; a lesson that restates a principle already there.

## Verification
- Lint and suites pass on the branch; the diff was shown and approved; `bin/learn count` dropped by the number handled; the ledger records each outcome.

Versions

The database has no other skills named dev-learn.

Database snapshot: 13.11.2026 · package version 5b9b3f0bf912 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100learn
    Extract key learnings from the current chat thread and save them to docs/learnings.md. Use this skill when the user types "/learn" or asks to save learnings from the conversation.
  • 86team-shinchan:learn
    Use when you want to manually add new learnings or rules to memory.
  • 100compound-learnings
    Transform session learnings into permanent capabilities (skills, rules, agents). Use when asked to "improve setup", "learn from sessions", "compound learnings", or "what patterns should become skills".
  • 100learn
    Start autonomous knowledge building daemon — browse learnings, store findings, synthesize. Use when user wants to learn, build knowledge graph, or grow expertise.
  • 100learn
    Capture session learnings and decisions to .suiperpower/learnings.md. Use when the user wants to save what they learned, log decisions, or wrap up a session.
  • 95learn
    Manage project learnings. Search, list, delete, promote, and export what idstack learned across sessions. Use when asked to "what have we learned", "show learnings", "prune stale learnings", or "export learnings". (idstack)

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.