Audit logs
Use when auditing and fixing logging in a scope — missing observability, INFO bloat that doesn't scale, stale or wrong-level messages, payload dumps on routine paths. Triggers on "audit logs", "fix logging", "logging review", "clean up log noise".From its SKILL.md
npx -y skills add JHostalek/dotclaude --skill audit-logsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 10 stars10 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
- runs commandsInstructs the agent to run 1 command, including `cat "${CLAUDE_SKILL_DIR}/../audit-workflow.md"`.
SKILL.md
1.4 KB, 249 tokens by cl100k_base, as published. Nobody here has run it
!cat "${CLAUDE_SKILL_DIR}/../audit-workflow.md"
Run as the logs dimension.
Three axes: missing, bloat, wrong. A reader reconstructing a production run from logs alone should see state changes, stage outcomes, and errors w/ enough context to debug — nothing else. INFO is what production pays for and scales w/ traffic; DEBUG is local-dev.
Missing: long-running stages w/o completion line; fan-outs w/o per-iteration progress at DEBUG; exceptions logged w/o stack; invisible state transitions.
Bloat: entry+exit duplicates; per-iteration INFO; timing scaffolding when trace store already records it; heartbeats on every poll regardless of outcome; payload dumps on routine paths.
Wrong: f-strings breaking aggregation; wrong level; field-name drift across same concept; stale templates no longer matching code.
Prefer demotion over deletion for INFO bloat — restoration harder than demotion (auto-fix). Exception: PII, session tokens, credentials in log fields → redact or hash, not demote — DEBUG still leaks in dev (sign-off). Architectural changes (wiring logging library, adopting trace store, restructuring routing) → sign-off.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.