メニュー
Image-quality gating and single-annotation tooling for captured screenshots (img-gate.mjs, annotate.mjs). Placeholder skill front door — T21.2 scope only.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
SOC 職業分類に基づく
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/bpmforge/attest --skill user-guideコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SKILL.md を表示中
Doc-driven implementation engineer. Reads SDLC design docs first, verifies all library APIs via Context7, then writes clean production code. Enforces anti-slop rules — no over-engineering, no defensive bloat, no hallucinated APIs. Use after design docs exist.
Gauntlet loop — multi-agent quality harness: a lead sets a real reference bar (a named product, test suite, or baseline to match or beat), splits the goal into gradeable units, builders produce artifacts in clean context, and blind fresh-per-round critics grade each artifact against the bar with evidence; failures loop until every unit passes, two rounds stall, or budget runs out. NOT /challenge (verifies factual claims); NOT /review (one-pass verdict) — this rebuilds until the work beats something real.
Claude-Design-style visual iteration loop — render the running UI, screenshot at 375/768/1440, critique against docs/design/tokens.json and design principles, apply fixes, re-capture until clean (cap 3 iterations). Also extracts a token baseline from an existing codebase (--sync) and audits real logged-in browsers (--real). NOT a one-pass review — use /ux --review for findings-only; NOT functional conformance — use /ui-verify.
| name | user-guide |
| description | Image-quality gating and single-annotation tooling for captured screenshots (img-gate.mjs, annotate.mjs). Placeholder skill front door — T21.2 scope only. |
Two Node.js tooling scripts built for the M21 user-guide capture pipeline
live at scripts/lib/ (relocated from this directory so they ship in BOTH
the OpenCode install and the generated Claude target — skills/*/scripts/
ships to neither):
scripts/lib/img-gate.mjs — Gate A quality check for a captured screenshot:
size floor, per-channel-stddev blank-detect, dominant-color-ratio vs a
per-app calibrated baseline, and two-shot pixel-diff stability. Returns a
pass/fail result with a specific reason per failed check.scripts/lib/annotate.mjs — draws one rounded highlight box and a numbered
badge onto a copy of a screenshot at a recorded bounding box; the
original is never mutated.Both require sharp (+ pixelmatch for img-gate) — direct deps of this
repo; installed projects need them available (npm i -D sharp pixelmatch).
See README.md in this directory for the per-app baseline calibration file
schema and usage examples of both scripts.
This SKILL.md exists only to satisfy this repo's npm test invariant
that every directory under skills/ has one — it is not the full
user-guide skill. The real skill front door (activation modes, coverage
reporting, assembly wiring into a manual) is a separate, later ticket that
authors this file for real, in both this repo and its generated sibling.
Until then, treat this file as a stub pointing at the two scripts above.