Menü
Audit the repository for common issues across detected technologies. Reports findings without fixing — the user decides what to act on.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Scan project to generate CLAUDE.md context files. Synthesizes overview from README, detects build commands, conventions, and references. Confirms findings with the user before writing. Re-running preserves user-confirmed and agent-discovered content.
Ensure the configured plans and memory directories exist, sync the plan format guide and review checklist, and archive completed plans older than 14 days. Run once before planning begins.
Basierend auf der SOC-Berufsklassifikation
| name | project-sanity |
| description | Audit the repository for common issues across detected technologies. Reports findings without fixing — the user decides what to act on. |
Audit the repository for common issues by detecting which technologies are present, running the corresponding sanity checks, and presenting all findings to the user.
This skill is audit-only — it reports findings but does not fix anything. The user decides what to act on after seeing the results.
Scan the repository root for these indicators and map each to a check file:
| Indicator | Check File |
|---|---|
.github/ directory exists | github-sanity.md |
codecov.yml or codecov.yaml exists at repo root | codecov-sanity.md |
Detect technologies — scan the repository root for each indicator in the detection table. Record which indicators are present and which check files to run.
Run check files — for each detected check file, read
.claude/skills/project-sanity/<check-file> and execute
its instructions. Collect all findings it reports.
Aggregate findings — combine findings from all executed check files into a single list. Preserve the source check file for each finding so the user knows which technology each finding belongs to.
Present findings — report all findings to the user, grouped by technology. For each finding include what was found and why it matters. Then ask the user what should be done about each finding.
