Parse audit-log.jsonl to surface guard verdicts, rewrite actions, and injection flags.
npx skills add https://github.com/Cain-Ish/claude-code-plugin --skill audit-cain-ish
Please help me install this Agent Skill. Skill: audit Source: https://github.com/Cain-Ish/claude-code-plugin/tree/main/skills/audit Command: npx skills add https://github.com/Cain-Ish/claude-code-plugin --skill audit-cain-ish
It helps you understand and verify what the second-brain safety layer did during a session by presenting the recorded guard verdicts and any flagged tool-return injections in a human-readable way.
Use the audit skill to review the last 100 safety-layer events for your current session by asking for an audit report for the current session with default settings.
High-intent search queries and answers about installing and using this skill.
Audit logging records PreToolUse guard verdicts and tool-return injection flags into a JSONL file. It captures session-level safety decisions, including denied actions, rewrites, and flagged injections for post-session investigation.
Filter audit logs by applying predicates on session_id, verdict, and hook fields within the JSONL records. This isolates specific safety events for focused troubleshooting and verification of guardrail behavior during incident investigation.
Yes, read-only auditing parses the JSONL log file without modifying state. It ensures trustworthy review of guard activity and tool-return scanner decisions after a session concludes, without fabricating results or altering system state.
Summarize denied and flagged safety events by parsing the JSONL audit log to aggregate guard verdict counts. This highlights rewrite actions and tool-return injection flags tied to specific hooks, providing clear session-level safety transparency.
Yes, the audit logging skill supports narrowing results by recency and hook name. You can focus your investigation by applying filters to the JSONL records, quickly finding relevant evidence tied to specific PreToolUse guards or verdicts.
Cain-IshScaffold Second Brain hot-tier files and register projects in a JSONL registry.
GRCEngClubAnalyze vendor security questionnaire responses for red flags, gaps, and follow-up needs.
Yapie0Blocks dangerous operations and redacts API keys from AI tool outputs.
sarathi-aimlSearch and analyze session JSONL logs for historical context.
