transcript-audit

Analyze agent-user conversation transcripts to identify skill gaps and failure patterns.

Updated Apr 3, 2026

Search Tags:#debugging #feedback #optimization #analysis #agent #transcripts

Ready to run

npx skills add https://github.com/strawgate/agent-skills --skill transcript-audit

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: transcript-audit
Source: https://github.com/strawgate/agent-skills/tree/main/transcript-audit
Command: npx skills add https://github.com/strawgate/agent-skills --skill transcript-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python3, and includes scripts (resource) components.

What problem does it solve?

This skill addresses the gap between how AI agents are currently performing and how users actually interact with them, identifying patterns of confusion, failure, and unmet needs.

Core Features & Use Cases

Transcript Discovery: Automatically locates and extracts conversation logs from various AI coding tools like OpenCode, Cline, and Claude Desktop.
Pattern Analysis: Identifies recurring task categories, context gaps, and specific failure signals in agent-user interactions.
Actionable Improvement: Provides a structured framework to propose concrete changes to existing skill prompts and logic based on real-world usage data.

Quick Start

Use the transcript-audit skill to find all local agent transcripts and generate a summary of the top task categories and missing capabilities.

Frequently Asked Questions about transcript-audit

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

FAQPage Schema

How do I analyze agent transcripts to find failure patterns and skill gaps?▼

To analyze agent transcripts, extraction scripts process JSON and JSONL conversation logs to identify recurring task categories, context gaps, and specific failure signals in agent-user interactions. This yields actionable improvement recommendations for existing skill prompts.

Can I extract conversation logs from tools like OpenCode, Cline, and Claude Desktop?▼

Yes, transcript discovery automatically locates and extracts conversation logs from various AI coding tools like OpenCode, Cline, and Claude Desktop. The extraction scripts process these local JSON and JSONL logs into evaluated interaction data.

Do I need Python 3 to extract and evaluate agent interaction data?▼

Yes, you need Python 3 installed to run the Python-based extraction scripts. These scripts are required to process JSON and JSONL logs into actionable improvement recommendations for your AI coding agents.

What is the best way to optimize AI agent performance using historical conversation data?▼

The best way to optimize AI agent performance is to audit historical conversation transcripts to identify patterns of confusion and unmet needs. This provides a structured framework to propose concrete changes to existing skill prompts and logic.

How does transcript analysis propose changes to existing skill prompts?▼

Transcript analysis proposes changes to skill prompts by evaluating real-world usage data from agent-user interactions. It identifies recurring task categories and missing capabilities, providing a structured framework to update skill logic accordingly.

What formats do the extraction scripts support for processing agent transcripts?▼

The extraction scripts support processing JSON and JSONL log formats. They extract and sample interaction data from these file formats to identify skill gaps and failure patterns across multiple AI coding tool environments.