From snoop
Analyzes Claude Code transcripts to generate post-mortem review reports with executive summaries, key findings by severity, and actionable recommendations. Use for reviewing, debugging, or auditing past sessions.
How this skill is triggered — by the user, by Claude, or both
Slash command
/snoop:reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Analyze a run transcript to identify issues and improvement opportunities.
Analyze a run transcript to identify issues and improvement opportunities.
/snoop:review # Analyze latest transcript
/snoop:review abc123 # Analyze specific transcript by ID
/snoop:review abc123 token usage # Analyze with focus on token usage
Arguments: $ARGUMENTS
| Input Format | Resolution |
|---|---|
abc123.jsonl | Use as-is |
abc123 (8 chars) | Append .jsonl |
| No identifier | Read from latest pointer |
Remaining text after identifier is the concern to focus on.
If identifier provided: Use .claude/transcripts/{identifier}.jsonl
If no identifier: Read from .claude/transcripts/latest pointer file, or fall back to:
ls -t .claude/transcripts/*.jsonl 2>/dev/null | head -1
If no transcript files exist, report to user and stop.
Delegate to the transcript-reviewer agent with the resolved file path:
Use the transcript-reviewer agent to analyze {resolved_path}
If a concern was extracted from arguments, include it:
Use the transcript-reviewer agent to analyze {resolved_path}. Focus on: {concern}
Present the review report. Include:
If critical issues found, highlight them prominently.
npx claudepluginhub rcrsr/claude-plugins --plugin snoopCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.