skills/scouting-anomalies/SKILL.md
File-by-file read-through diagnostic to surface anomalies that fixed checklists miss. Use when: 通読, 違和感, 読み通し, 気になるとこ, read-through, anomaly scout, file-by-file review. Do NOT use for full audit (use /audit), PR screening (use /preview), or readability review (use reviewing-readability).
npx skillsauth add thkt/claude-config scouting-anomaliesInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Read-through diagnostic. Staging行き、audit直結しない。
| Step | Action |
| ---- | ----------------------------------------------------------------------------- |
| 1 | 対象ファイルを 1 つずつ通読 (Read, ±20 lines) |
| 2 | 違和感候補を文字化 (1 行要約 + file:line) |
| 3 | Filter 適用 (下表) |
| 4 | 該当 reviewer の ID prefix 配列 + 該当 Category を分類 |
| 5 | staging 追記 ($HOME/.claude/workspace/scout/staging.md。無ければ mkdir -p) |
Yes/YesのみSTAGE、それ以外SKIP。
| Filter | Question | | --------------- | --------------------------------------------------------------- | | Senior Engineer | Would a senior engineer request a change? | | Harm | Concrete trigger for bug/data loss/security/maintenance burden? |
正本: ../../templates/scout/read-through-staging.md
| Field | Value |
| --------------------- | --------------------------------------------------------------------------------------- |
| severity | critical / high / medium / low |
| evidence | file:line + observation |
| reasoning | concrete language |
| existing-reviewer-fit | ID prefix 配列 (例: [CQ, PQ])。該当なしは [] |
| category-in-reviewer | 該当 reviewer の Category (例: PQ の clarity、CQ の readability)。該当なしは none |
ID prefixとCategoryは ../../templates/audit/finding-schema.md のRegistryと各reviewer定義を参照。Language Constraintsも準拠。
tools
Delegate implementation to codex (coder) via the herdr-agentchat plugin and drive a two-pane conversation to completion.
development
Extract recurring patterns from past closed PRs/issues and the research findings in workspace/research/, verify them against the latest code, and propose them to docs/wiki/ via PR.
development
Create Decision Records (DR) in MADR v4 format with auto-numbering.
development
Codex review + cleanup. Findings are challenged by critic-audit, not aggregated as facts, and fixes are applied directly. Do NOT use for internal multi-reviewer deep audits or findings reports (use /audit).