メニュー
Generate semantic commit messages from staged changes. Use when committing code to produce consistent, well-structured commit messages following conventional commit format.
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。
SOC 職業分類に基づく
Prepare GitHub pull requests for review and follow up on PR review comments. Use before any PR creation or update, including "open a PR", "create a PR", "make a pull request", `gh pr create`, writing/updating a PR body, requesting review, handling Copilot or human review feedback, resolving comments, or asking whether a branch is ready to review. Includes rubber ducking, semantic commits, PR body/review guidance, 5-minute Copilot feedback follow-up scheduling, and work-repo vs personal-project boundaries.
Reviews PRs, local diffs, and code-adjacent design changes from a principal engineer standpoint. Use for high-signal review, MVP scope control, maintainability, security-aware review, reuse of existing primitives, avoiding premature optimization, and deciding what must change before merge. When a review turns into architecture/API/platform design and named laws would sharpen the critique, compose with software-design-laws as an optional lens.
Prevents unnecessary dependency installation and audits unavoidable packages. Use before adding, installing, upgrading, or recommending dependencies, packages, libraries, gems, npm modules, pip packages, crates, GitHub Actions, CLIs, or package-manager commands.
Generate images locally using Ollama's x/z-image-turbo model. Use when the user asks to generate an image, create art, make a picture, or produce a visual. Trigger phrases include "generate an image", "create a picture", "make art", "image of", "draw me", "visualize", "photo of", "illustration of".
Stress-test plans, designs, proposals, and implementation strategies by interviewing the user one decision at a time. Use when the user says "stress test this", "interview me", "improve this plan", "challenge this plan", "poke holes in this", or wants plan refinement.
Run a multi-model roundtable for architecture decisions, ambiguous tradeoffs, strategy questions, investigation framing, or when the user asks for multiple model perspectives. Fans prompts out to several models, critiques disagreement, and synthesizes consensus plus dissent. Not for fleet mode.
| name | semantic-commit |
| description | Generate semantic commit messages from staged changes. Use when committing code to produce consistent, well-structured commit messages following conventional commit format. |
Generate commit messages following the conventional commit format: <type>(<scope>): <subject>
<type>(<scope>): <subject>
<body>
| Type | Use for |
|---|---|
feat | New user-facing feature |
fix | Bug fix |
docs | Documentation only |
style | Formatting, whitespace (no logic change) |
refactor | Code change that doesn't fix or add |
test | Adding or updating tests |
chore | Build, CI, dependencies |
Trivial:
docs: fix typo in readme
Simple with scope:
feat(auth): add password reset flow
With body:
fix(api): handle empty response from upstream
Add null check before parsing response body. Previously, empty
responses from the payments service caused JSON parse errors.
git diff --cached