atelier-fashion/adlc-toolkit/tasks

Claude Code and agent skill by atelier-fashion/adlc-toolkit

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Install

npx vskill@latest install atelier-fashion/adlc-toolkit/tasks

SKILL.md

Description

Create manifest/SKILL.md — a standalone, read-only skill that derives a cross-session view of in-flight ADLC work from the remote (open PRs + pushed feat/REQ-* branches), enriches each entry with component/domain, computes a coarse overlap report, and renders an advisory table. This is the foundational artifact; the wiring tasks (066–068) depend on it.

Files to Create/Modify

  • manifest/SKILL.md — the new skill (create)

Acceptance Criteria

  • [ ] Skill file follows the canonical skeleton: frontmatter (name: manifest, description, argument-hint), # /manifest — … title, ## Ethos injection macro (!sh .adlc/partials/ethos-include.sh 2>/dev/null || sh ~/.claude/skills/partials/ethos-include.sh`), ## Context, ## Input (optional REQ filter / self), ## Prerequisites (git repo + reachable origin; fail clearly otherwise), ## Instructions, ## Quality Checklist`.
  • [ ] Enumerates in-flight REQs from BOTH gh pr list --state open --json number,headRefName,author,isDraft,createdAt,url --limit 200 (incl. drafts) AND remote feat/REQ-* branches (git branch -r --list 'origin/feat/REQ-*'), deduped by REQ id (PR entry wins; branch-only added when no PR). (BR-3, BR-4)
  • [ ] Maps branch → REQ via feat/REQ-<digits>-<slug>; non-matching branches are silently ignored. (BR-4)
  • [ ] Every remote-derived identifier is sanitized before shell use: REQ ids validated against ^REQ-[0-9]{3,6}$; substituted branch/path values single-quoted; any path segment equal to .. rejected before git show. (BR-5, LESSON-008)
  • [ ] Component/domain enrichment tries local frontmatter first, then git show origin/<branch>:.adlc/specs/<REQ>-*/requirement.md, then unknown; an entry is never dropped on enrichment failure. (BR-11, BR-12)
  • [ ] Coarse overlap report flags pairs sharing component OR domain (labels which), presented as ADVISORY only with an explicit "no action enforced" note. (BR-8)
  • [ ] Includes the current session's own REQ when one is in context, marked as self. (BR-13)
  • [ ] Graceful degradation: with gh unavailable/unauthenticated, enumerate from branches only, annotate that PR data is missing, exit 0 — never hard-fail. (BR-6)
  • [ ] Strictly read-only: no branch/PR/file/state mutation; git status clean after a run. (BR-2) No stored manifest file is created. (BR-1)
  • [ ] Network cost is O(1) in the number of in-flight REQs: reuse caller's fetch in pre-flight contexts / one own git fetch standalone, one batched gh pr list, local git show enrichment — no per-branch API calls. (BR-14)
  • [ ] Output table columns: REQ | author | branch/PR | state | component/domain | opened-at (+ self marker). (AC from requirement)
  • [ ] python3 tools/lint-skills/check.py passes clean (no new findings).
  • [ ] Dogfood: running /manifest in this repo lists feat/REQ-482-manifest-skill as in-flight (self) and exits 0.

Technical Notes

  • Mirror status/SKILL.md for structure and table style; mirror proceed/SKILL.md / template-drift/SKILL.md for the identifier-sanitization + path-traversal-rejection patterns.
  • No Kimi (ADR-3): do NOT reference ADLC_DISABLE_KIMI, kimi-gate.sh, or telemetry — this skill does git/gh I/O, not bulk reads, and referencing the gate would trip the lint canonical-helper check.
  • POSIX/BSD bash only (conventions "Bash in skills"): grep -oE not -oP; no local inside `sh / `shell fences (use `bash if local is truly needed); keep any shell function defined-and-invoked within the same fenced block (cross-fence-fn lint check — LESSON-020).
  • Keep $(/) and $((/)) balanced within each fence (lint shell-balance check).
  • state values: draft (PR isDraft) | ready (open non-draft PR) | no-pr (branch only).

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
5

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/atelier-fashion/adlc-toolkit/tasks/badge)](https://verified-skill.com/skills/atelier-fashion/adlc-toolkit/tasks)

Something wrong with this skill or its page? Tell us.