メニュー
Research current best practices and turn the evidence into a clear recommendation. Use for state-of-the-art, evidence-based, optimal-approach, comparison, or pros-and-cons questions in any field.
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
Compare Claude Code and Codex on the same real code-change task with isolated worktrees, identical gates, transcripts, time, and cost.
Align AGENTS.md, rules, README, plans, and chronicles with the codebase; use --clean to archive obsolete task docs into ATLAS.md.
Clarify an ambiguous or consequential change before planning or implementation when more than one sound approach exists.
Add CHANGELOG.md entries, derive them from commits, or cut a Keep a Changelog and SemVer release.
Create a user-requested Conventional Commit from the staged changes.
Define or implement regression proof for a project or business flow. Use for test strategy, black-box or deep-integration tests, business KPIs and thresholds, test audits, missing tests, or test implementation.
| name | best-practices |
| description | Research current best practices and turn the evidence into a clear recommendation. Use for state-of-the-art, evidence-based, optimal-approach, comparison, or pros-and-cons questions in any field. |
| argument-hint | <topic> |
| user-invocable | true |
| allowed-tools | WebSearch, WebFetch, Read, Glob, Grep, Skill |
Match the research effort to the decision: use a quick lookup for a narrow fact; use deep, multi-source research for costly, risky, disputed, or fast-changing choices.
On Claude, use /deep-research when available. On Codex, perform the equivalent iterative web research. Otherwise use the platform's best search and browsing tools.
Prefer primary sources, cite material claims inline, and separate evidence from inference. Explain consensus, disagreement, uncertainty, and where each recommendation applies. Recommend for the user's goals and constraints, not for an imaginary average user.
If missing context changes the answer, ask for it or give clearly conditional recommendations. Do not pad the report with source quotas, evidence labels, or weak secondary material.
