agentleFS
blueclover22blueclover22/mw_agent_kit/skills/dev-resume/SKILL.md

dev-resume

Skill1 starsChanged 8 days agoMITView on GitHub

Use only on explicit request to resume or take over work when the next task is undecided. Use mak:dev-kickoff for a decided task, mak:reverse-engineering when project docs are absent, and mak:doc-audit for document-consistency audits.

The file

Files
SKILL.md

mak:dev-resume — Work Re-Entry Point

Derives, from what project documents already show, (a) how far work has progressed, (b) problems in that state, (c) one concrete next step — each backed by evidence. Owns the inbound side of session handoff: finding out what to do, not doing it.

Report and all user-facing output in the user's conversation language (or the project's documented language policy).

Write boundary

This skill diagnoses, reports, and routes only — fixes go through the normal flow. Allowed: Read / Grep / Glob, read-only git commands, and reproducing only the verification commands the documents actually claim (§Procedure step 3). Forbidden: creating/modifying/deleting any file and any code change; found defects go on the next-step candidate list.

When to Use

Resuming in-progress work, or picking up a handed-off session, when the next concrete task itself is undecided.

Route elsewhere:

SituationGo to
No project documents at allmak:reverse-engineering
Document drift dominates over "what's next"mak:doc-audit
The roadmap/phase structure itself is stalemak:roadmap-planning
The task is already decidedmak:dev-kickoff

Scope Rule

Fixed entry points only — inherits mak:doc-audit §Scope Rule; never a full codebase scan.

Entry points: the roadmap's SSOT status (located per the mak:roadmap-planning save-path priority), design docs (located per the mak:design-doc-template save-path rule) — their meta Status values and §5.0 Step → verify status column, recent commits, uncommitted changes — plus the paths/symbols those documents cite.

Procedure

  1. Confirm current position — read the entry points above. The design doc's §5.0 status column is the stated progress: read it first and treat it as a claim, not as truth. Infer position from commits and code only where that column is missing, or where step 3 shows it contradicts the observed state
  2. Collect entry points — trace cited paths/symbols per §Scope Rule
  3. Verify claims — reproduce only commands that verify a claim the documents actually make (stated test counts, "clean" lint/build, etc.); read only cited code. No claim → no command
  4. Report diagnosis — see §Report
  5. Present next steps — 2–3 candidates, ask the user to choose. If a document already designates a next item, it must be one of the candidates

Report

Reuse mak:review-report grades — 🔴 Critical / 🟡 Warning / 🟢 Pass / 📝 Note — no new grades. Every finding pairs a document's claim with the observed fact. Distinguish "no issue found" from "could not verify".

GradeItemLocationEvidence (claim vs. observed)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.