transcript-condenser

Condenses Claude Code session transcripts into searchable, structured summaries.

3|Updated Oct 6, 2025
One-click install
npx skills add https://github.com/nitromike502/claude-code-web-manager --skill transcript-condenser
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: transcript-condenser
Source: https://github.com/nitromike502/claude-code-web-manager/tree/main/.claude/skills/transcript-condenser
Command: npx skills add https://github.com/nitromike502/claude-code-web-manager --skill transcript-condenser

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Claude Code session transcripts can be verbose and difficult to parse, making it hard to quickly review past interactions, understand tool usage, or analyze workflow patterns. This Skill automates the condensation of these logs, transforming raw data into actionable insights.

Core Features & Use Cases

  • Human-Readable Summaries: Condenses lengthy transcripts into clear Markdown or structured JSON, filtering out noise like metadata and system messages.
  • Workflow Analysis: Easily identify tool usage patterns, subagent interactions, and key events to understand how tasks were executed.
  • Customizable Output: Choose verbosity levels (minimal, standard, detailed) and filter options (tool-only, subagent-only) to focus on what matters most.
  • Batch Processing: Process entire directories of transcripts, generating consistent reports for multiple sessions.
  • Use Case: Quickly review your last 5 Claude Code sessions to see which tools were most effective, identify common errors, and pinpoint areas for workflow optimization.

Quick Start

Condense a single transcript to markdown (stdout)

node .claude/skills/transcript-condenser/condense-transcript.js <transcript-file>

Save condensed version to file

node .claude/skills/transcript-condenser/condense-transcript.js <transcript-file> --output=summary.md

Frequently Asked Questions about transcript-condenser

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

FAQPage Schema
How do I summarize Claude Code session transcripts?

Summarize transcripts by running the condense-transcript script on your session files. It transforms verbose logs into readable Markdown or JSON, filtering out metadata and system messages to reveal the core interactions and tool usage patterns from your Claude Code work.

Can I filter Claude Code transcripts to show only tool usage?

Yes, the transcript condenser supports filtering options including tool-only and subagent-only modes. You can also adjust verbosity levels (minimal, standard, detailed) to focus on specific aspects of your session, making it easier to analyze workflow patterns.

How do I batch process multiple Claude Code session transcripts?

Batch processing lets you condense entire transcript directories at once, generating consistent reports across multiple sessions. The script applies your chosen output format and verbosity settings uniformly, saving time when reviewing a collection of Claude Code interactions.

What output formats does the transcript condenser support?

The transcript condenser outputs in Markdown (human-readable summaries) or JSON (structured data). You can pipe results to stdout or save directly to a file, making it flexible for integration into documentation, analysis workflows, or archival systems.

When should I use transcript condensing for Claude Code sessions?

Condensing is useful when you need to quickly review past sessions, identify which tools were most effective, spot common errors, or understand workflow patterns without parsing raw, verbose logs manually.