Verified:GitHub

transcript-analysis

Analyze Claude Code transcript logs to quantify model routing and debugging convergence.

4|1|Updated Mar 11, 2026

Search Tags:#pull requests #test failures #transcript analysis #model routing #code debugging #review enforcement #session timelines

Ready to run

npx skills add https://github.com/jcdendrite/claude-config --skill transcript-analysis-jcdendrite

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: transcript-analysis
Source: https://github.com/jcdendrite/claude-config/tree/main/claude/.claude/skills/transcript-analysis
Command: npx skills add https://github.com/jcdendrite/claude-config --skill transcript-analysis-jcdendrite

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Transcript analysis helps you understand why Claude Code sessions succeed or fail by turning messy conversation logs into measurable patterns like convergence vs thrashing, reviewer activity, and model routing behavior.

Core Features & Use Cases

Branch- and session-level diagnostics: Identify which branches exist and which models ran on each, then measure failure convergence sequences, struggle signals, and active-vs-idle timing.
Debugging loop shaping: Determine whether test debugging tends to converge (fix lands) or thrash (no sustained progress), using sequence signatures and streak metrics.
PR and review trace mapping: Link branches to PRs and count review comments, then locate sessions that triggered review skills, enforcement hook denials, or reviewer-agent spawns.

Quick Start

Run transcript-analysis to analyze your local Claude Code transcript data using the toolkit at ~/.claude/scripts/transcript-analysis.py.

Frequently Asked Questions about transcript-analysis

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

FAQPage Schema

How do I analyze Claude Code transcript logs to find test failure patterns?▼

Transcript analysis turns conversation logs into measurable patterns by computing metrics from logged runs without external writes. It quantifies model routing, debugging convergence, and review enforcement outcomes across branches and sessions.

How do I track model routing and subagent time split in Claude Code sessions?▼

You can map PRs to branches and locate review enforcement sessions by running the transcript-analysis tool. It links branches to pull requests, counts review comments, and traces sessions that triggered enforcement hook denials or reviewer-agent spawns.

Does transcript analysis require external dependencies or write access to work?▼

Transcript analysis requires no external dependencies and performs no external writes. It operates entirely locally by calling the transcript-analysis.py tool to compute metrics from your existing Claude Code logged runs within a selected date window.

What are the limitations of analyzing conversation logs for debugging convergence?▼

The limitation is that analysis is restricted to computing metrics from local logged runs within a specified date window. It requires the transcript-analysis.py tool to be present locally and cannot perform external writes or fetch missing session data.