ci

スター1

フォーク0

更新日2026年5月14日 22:34

CI/CD workflow rules — GitHub Actions patterns, validation, Playwright in CI. Auto-invoke when editing .github/workflows files or discussing CI pipelines.

インストール

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。

ソース

fubits1

fubits1/ronin-skills

GitHub リポジトリを開く Creator のリポジトリを見る

ダウンロード

CI

ALWAYS validate workflow files with pnpx node-actionlint <file> after any edit. Exit code 0 or it's not valid.
Playwright in CI: No official setup action (deprecated). Use pnpm exec playwright install --with-deps chromium to install browser + OS-level system dependencies. Only install the browsers you actually need.
Simulate CI locally before pushing: Run commands with CI=true GITHUB_ACTIONS=true to catch tools that auto-detect CI and change behavior.
Single validation entry point for CI. Define one script (e.g. validate) that chains lint + typecheck + unit tests + framework-check. CI calls only that.
Tag-filtered CI test command. For test suites where some cases require a live backend or recorded fixtures unsuitable for CI, define a CI-specific script that filters those out (e.g. via tag filter --tags-filter='!ci-skip'). Keeps CI deterministic.
Separate concerns by script. test:unit (unit only), test:e2e (browser/e2e only), test (everything concurrently for local), test:ci (the filtered subset). Don't overload one script with multiple modes.
Check the exact commit SHA that triggered a CI run. Don't carry stale assumptions about what code is running.
Read ALL errors in CI logs. Download the full log with gh api repos/.../actions/jobs/<id>/logs, then grep for FAIL, Error, unhandled, ECONNREFUSED. CI failures often have MULTIPLE root causes — fix all of them, not just the first one.
pnpm test -- --flag does NOT work for passing flags through pnpm to the underlying test runner. The -- makes the runner treat flags as positional args. Define a dedicated script (e.g. test:ci) instead of trying to pass flags through.
continue-on-error pattern: use when you want to collect results from multiple steps and report them together (e.g. PR comment tables) instead of failing fast.
Adding a build step is NOT free. Estimate build time vs time saved before adding one. A 3min build for marginally faster page serving = net loss.

このリポジトリの他の Skills

同じリポジトリ

no-bash

fubits1/ronin-skills

Use fff MCP / Grep / Read / Glob — never Bash grep/cat/find/head/tail/sed/awk/rg/wc — for file search and read. Use jq for JSON parsing/shaping (not node -e / python -c). A hook hard-blocks the banned Bash calls, and the dedicated tools return clean, cacheable, line-numbered results. Auto-invoke when searching, reading, or counting in files.

2026-06-291

plan

fubits1/ronin-skills

Planning and problem-solving discipline: enter plan mode first, write plans in the plan file (not chat), structured research, validation, implementation. Auto-invoke when the user says "plan", "make a plan", or when fixing issues or implementing features.

2026-06-211

before-you-act

fubits1/ronin-skills

Use before any action: five-question self-check that catches destructive ops, scope creep, fabrication, premature completion, and ignored instructions. Auto-invoke before executing commands, editing files, or answering factual questions.

2026-06-201

update-skills

fubits1/ronin-skills

Reconcile skill collections, backport from ~/.claude/skills into a marketplace plugins tree, scrubbing proprietary refs. Use when local has diverged from upstream, or when promoting local-only skills to a marketplace.

2026-06-201

discipline

fubits1/ronin-skills

Communication and scope discipline: how to interact with the user, handle rejection, stay in scope. Always active.

2026-06-191

done

fubits1/ronin-skills

Final checklist before declaring any task complete. Auto-invoke before saying "done", "ready", "complete", or asking to commit/push.

2026-06-181

ALWAYS validate workflow files with pnpx node-actionlint <file> after any edit. Exit code 0 or it's not valid.

Playwright in CI: No official setup action (deprecated). Use pnpm exec playwright install --with-deps chromium to install browser + OS-level system dependencies. Only install the browsers you actually need.

Simulate CI locally before pushing: Run commands with CI=true GITHUB_ACTIONS=true to catch tools that auto-detect CI and change behavior.

Single validation entry point for CI. Define one script (e.g. validate) that chains lint + typecheck + unit tests + framework-check. CI calls only that.

Tag-filtered CI test command. For test suites where some cases require a live backend or recorded fixtures unsuitable for CI, define a CI-specific script that filters those out (e.g. via tag filter --tags-filter='!ci-skip'). Keeps CI deterministic.

Separate concerns by script. test:unit (unit only), test:e2e (browser/e2e only), test (everything concurrently for local), test:ci (the filtered subset). Don't overload one script with multiple modes.

Check the exact commit SHA that triggered a CI run. Don't carry stale assumptions about what code is running.

Read ALL errors in CI logs. Download the full log with gh api repos/.../actions/jobs/<id>/logs, then grep for FAIL, Error, unhandled, ECONNREFUSED. CI failures often have MULTIPLE root causes — fix all of them, not just the first one.

pnpm test -- --flag does NOT work for passing flags through pnpm to the underlying test runner. The -- makes the runner treat flags as positional args. Define a dedicated script (e.g. test:ci) instead of trying to pass flags through.

continue-on-error pattern: use when you want to collect results from multiple steps and report them together (e.g. PR comment tables) instead of failing fast.

Adding a build step is NOT free. Estimate build time vs time saved before adding one. A 3min build for marginally faster page serving = net loss.

name	ci
description	CI/CD workflow rules — GitHub Actions patterns, validation, Playwright in CI. Auto-invoke when editing .github/workflows files or discussing CI pipelines.
user-invocable	true