메뉴
Run tiered computational quality sensors and self-correct. Use after code changes, before commits/PRs, when asked to "verify", "run gates", "quality check", or when AGENTS.md requires verification. Prefer gate:fast during iteration and gate:full before handoff.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Inferential code review sensor for this repo. Use before asking a human to review, after verify gates are green, or when asked to "review", "review changes", or "pre-review". Checks harness compliance, architecture layers, test adequacy, and agent failure modes — not a substitute for cargo/clippy/CI.
Behaviour harness for the 8 ASCII Canvas drawing tools. Use when changing tools, canvas interaction, border styles, or when AGENTS.md tool checklist applies.
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
Write clear, comprehensive technical documentation. Use when creating specs, architecture docs, runbooks, or API documentation. Handles technical specifications, system design docs, operational guides, and developer documentation with industry best practices.
Generates GitHub Actions CI workflow configuration
| name | verify |
| description | Run tiered computational quality sensors and self-correct. Use after code changes, before commits/PRs, when asked to "verify", "run gates", "quality check", or when AGENTS.md requires verification. Prefer gate:fast during iteration and gate:full before handoff. |
Part of the project harness (agents-docs/harness.md). Runs deterministic sensors and drives the agent self-correction loop.
| Tier | Command | Use |
|---|---|---|
| fast | npm run gate:fast | Default after edits |
| full | npm run gate:full | Before PR |
| architecture only | ./scripts/check-architecture.sh | Layer/import changes |
| focused | cargo test … / cd web && pnpm test / one Playwright file | Tight loop |
npm run gate:fast
# or
npm run gate:full
[FAIL] and FIX: lines. Fix root cause. Re-run the same tier.#[allow], skip tests, or expand .loc-allowlist without an ADR.tool-validation skill / relevant E2E.code-review skill after full gates are green.-D warnings, build, cargo testweb/pkg exists (builds WASM if missing — pkg is gitignored)tsc --noEmit, Vitestpnpm run build:wasm + check-size (≤ 1.5MB) if not already built| CI job | Local equivalent | Gotcha |
|---|---|---|
| Web tsc | cd web && pnpm exec tsc --noEmit | Needs web/pkg (gitignored). Clean tree: npm run build:wasm first. See harness L-001. |
| Architecture | ./scripts/check-architecture.sh | Needs rg |
| WASM size | npm run check-size | After build:wasm |
| E2E | npx playwright test --project=chromium | Needs pkg + dev server |
If CI fails but local gates passed, treat it as a harness bug: update sensors so local fails the same way (append to agents-docs/harness.md Learned failure modes).
If you hit the same failure class twice in a session (or it recurred from a past PR):
AGENTS.md rule, sensor message, or CI job dependency.agents-docs/harness.md.plans/TECHNICAL_ANALYSIS.md when non-obvious.rust-engineer / typescript-expert → verifycode-review → human PRtool-validation, dogfood