tau_audit

Parse AI agent audit logs to identify bottlenecks, errors, and anomalies.

Updated Jun 22, 2026
One-click install
npx skills add https://github.com/alangeb/tau --skill tau-audit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tau_audit
Source: https://github.com/alangeb/tau/tree/main/src/skills/tau_audit
Command: npx skills add https://github.com/alangeb/tau --skill tau-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill solves the difficulty of debugging complex AI agent sessions by providing a structured, automated way to parse and analyze audit logs for errors, loops, and tool usage patterns.

Core Features & Use Cases

  • Automated Log Analysis: Quickly identify session health, tool failure rates, and loop detection across multiple log files.
  • Behavioral Insights: Extract user-to-assistant interaction patterns, fork/subagent nesting, and tool chain latency.
  • Use Case: When an agent enters an infinite loop or fails repeatedly, use this Skill to batch-analyze the logs to pinpoint the exact tool call or error type causing the breakdown.

Quick Start

Use the tau_audit skill to analyze the session log file located at .audit and generate a comprehensive health report.

Frequently Asked Questions about tau_audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug AI agent loops and repeated tool failures from session logs?

To debug AI agent loops, parse structured audit logs to detect loop occurrences, pinpoint failing tool calls, and identify the root cause of repeated execution errors.

What is the best way to analyze AI agent performance bottlenecks and tool latency?

Analyzing AI agent performance bottlenecks involves parsing audit logs to evaluate tool success rates, measure tool chain latency, and correlate call sequences with system warnings.

Can I batch process multiple AI agent session files to check overall health?

Yes, you can batch process multiple structured session files to evaluate overall session health, detect behavioral anomalies, and generate comprehensive health reports.

How do I identify behavioral anomalies in AI agent interactions?

To identify behavioral anomalies in AI agent interactions, parse audit logs to extract user-to-assistant patterns, monitor subagent nesting, and detect execution deviations.

Does AI agent log analysis work for evaluating context compression efficiency?

Yes, AI agent log analysis evaluates context compression efficiency by correlating structured audit log data with assistant response quality and system warnings during execution.

Why does my AI agent session break down during complex tool call sequences?

AI agent sessions break down when tool call sequences trigger execution errors or loops; root cause analysis correlates these sequences with response quality and system warnings to identify failures.