Convert Claude Code JSONL session files to readable markdown transcripts and share them with your friends.

uv tool install claude-transcript
cd ~/.claude/projects/your-project-path
claude-transcript {session-id}.jsonl --upload-gist

uv tool install claude-transcript

uv add claude-transcript
# or
pip install claude-transcript

# Basic usage
claude-transcript {session-id}.jsonl

# Custom output filename
claude-transcript {session-id}.jsonl -o custom_name.md

# Upload to GitHub Gist after generation
claude-transcript {session-id}.jsonl --upload-gist

# Show help
claude-transcript --help

Claude Code automatically saves all conversations locally in JSONL (JSON Lines) format. Each session creates a file containing structured data about your interaction:

• Storage Location: Conversations are stored locally on your machine with full message history in ~/.claude/projects/{project-path}
• File Format: JSONL files contain one JSON object per line, capturing messages, tool operations, metadata, and timing information
• Content Types: Files include user messages, assistant responses, tool executions (file edits, searches, bash commands), sidechains for parallel processing, and session metadata

Each line in a Claude Code JSONL file represents a different type of event or operation:

• Message exchanges between user and assistant
• Tool operations and their results
• Timing and cost information
• Parallel task execution data from sidechains

This structured format allows for comprehensive reconstruction of the entire conversation flow, which this tool converts into readable markdown transcripts.

• Zero dependencies: Pure Python with no external dependencies
• Comprehensive parsing: Handles all JSONL entry types including tool operations, sidechains, and timing information
• GitHub Gist integration: Optional upload to share transcripts easily
• Tool operations display: Shows detailed information about file edits, searches, and other actions taken during the session

The tool generates clean, readable markdown transcripts that include:

• Session metadata and timing information
• Conversational turns with user requests and assistant responses
• Detailed tool operation logs (file edits, searches, bash commands, etc.)
• Parallel task execution details from sidechains

• Python 3.8 or higher
• No external dependencies required
• Optional: GitHub CLI (gh) for gist upload functionality

This tool is built as a standalone script with PEP 723 inline dependencies, making it easy to run directly with uv run or package as a proper Python package.

MIT License - see LICENSE file for details.