extract-transcripts

0xBigBoss 的头像作者 0xBigBoss

从 Claude Code 和 Codex CLI 会话 JSONL 文件中提取可读的转录本。

📖 中文导读:从 Claude Code 和 Codex CLI 会话 JSONL 文件中提取可读的转录本。 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: extract-transcripts description: Extract readable transcripts from Claude Code and Codex CLI session JSONL files

Extract Transcripts

Extracts readable markdown transcripts from Claude Code and Codex CLI session JSONL files.

Scripts

Claude Code Sessions

# Extract a single session
uv run ~/.claude/skills/extract-transcripts/extract_transcript.py <session.jsonl>

# With tool calls and thinking blocks
uv run ~/.claude/skills/extract-transcripts/extract_transcript.py <session.jsonl> --include-tools --include-thinking

# Extract all sessions from a directory
uv run ~/.claude/skills/extract-transcripts/extract_transcript.py <directory> --all

# Output to file
uv run ~/.claude/skills/extract-transcripts/extract_transcript.py <session.jsonl> -o output.md

# Summary only (quick overview)
uv run ~/.claude/skills/extract-transcripts/extract_transcript.py <session.jsonl> --summary

# Skip empty/warmup-only sessions
uv run ~/.claude/skills/extract-transcripts/extract_transcript.py <directory> --all --skip-empty

Options:

  • --include-tools: Include tool calls and results
  • --include-thinking: Include Claude's thinking blocks
  • --all: Process all .jsonl files in directory
  • -o, --output: Output file path (default: stdout)
  • --summary: Only output brief summary
  • --skip-empty: Skip empty and warmup-only sessions
  • --min-messages N: Minimum messages for --skip-empty (default: 2)

Codex CLI Sessions

# Extract a Codex session
uv run ~/.claude/skills/extract-transcripts/extract_codex_transcript.py <session.jsonl>

# Extract from Codex history file
uv run ~/.claude/skills/extract-transcripts/extract_codex_transcript.py ~/.codex/history.jsonl --history

Session File Locations

Claude Code

  • Sessions: ~/.claude/projects/<project-path>/<session-id>.jsonl

Codex CLI

  • Sessions: ~/.codex/sessions/<session_id>/rollout.jsonl
  • History: ~/.codex/history.jsonl

DuckDB-Based Transcript Index

For querying across many sessions, use the DuckDB-based indexer:

# Index all sessions (incremental - only new/changed files)
uv run ~/.claude/skills/extract-transcripts/transcript_index.py index

# Force full reindex
uv run ~/.claude/skills/extract-transcripts/transcript_index.py index --full

# Limit number of files to process
uv run ~/.claude/skills/extract-transcripts/transcript_index.py index --limit 10

# List recent sessions
uv run ~/.claude/skills/extract-transcripts/transcript_index.py recent
uv run ~/.claude/skills/extract-transcripts/transcript_index.py recent --limit 20
uv run ~/.claude/skills/extract-transcripts/transcript_index.py recent --project myapp
uv run ~/.claude/skills/extract-transcripts/transcript_index.py recent --since 7d

# Search across sessions
uv run ~/.claude/skills/extract-transcripts/transcript_index.py search "error handling"
uv run ~/.claude/skills/extract-transcripts/transcript_index.py search "query" --cwd ~/myproject

# Show a session transcript
uv run ~/.claude/skills/extract-transcripts/transcript_index.py show <file_path>
uv run ~/.claude/skills/extract-transcripts/transcript_index.py show <file_path> --summary

Requirements: uv (dependencies auto-installed via inline script metadata)

Database location: ~/.claude/transcript-index/sessions.duckdb

Output Format

Transcripts are formatted as markdown with:

  • Session metadata (date, duration, model, working directory, git branch)
  • User messages prefixed with ## User
  • Assistant responses prefixed with ## Assistant
  • Tool calls in code blocks (if --include-tools)
  • Thinking in blockquotes (if --include-thinking)
  • Tool usage summary for Codex sessions