Claude Session Auditor FAQs

Question 1

What metrics does the audit provide?

Accepted Answer

It provides comprehensive data on token usage, turn counts, tool execution success rates, timing information, and specific line-by-line context for errors.

Question 2

What is the primary use of the session-audit skill?

Accepted Answer

The session-audit skill is used to review past Claude Code interactions, identify where the model made mistakes or required human redirection, and extract data to improve future performance.

Question 3

How do I find a specific session ID for auditing?

Accepted Answer

The skill includes a resolution script that automatically searches your local Claude project directories (~/.claude/projects/) to find the relevant JSONL transcript files by UUID.

Question 4

How does this help in refining AI skills?

Accepted Answer

By identifying 'floundering' points—where the agent retries tools or takes wrong paths—you can identify gaps in your skill's instructions and create more precise, direct prompts.

Question 5

Can it analyze subagents spawned by the main Claude instance?

Accepted Answer

Yes, the auditor identifies subagent sessions associated with a main session ID and can perform the same detailed error and stats analysis on those sub-tasks.

Claude Session Auditor

Key Features

Use Cases

Claude Session Auditor

Key Features

Use Cases