extract-transcripts

Extracting readable transcripts from Claude Code and Codex JSONL files involves parsing session data and converting it into Markdown. This process captures user and assistant messages alongside tool calls to create an easily viewable archive of your conversations.

The best way to search across multiple Claude Code session transcripts is by building a DuckDB-based index. This enables cross-session analysis and querying across large transcript collections, allowing you to quickly locate specific information within your archives.

You can convert Codex CLI session data into Markdown for archival by processing the JSONL files. This functionality supports auditing and knowledge capture by transforming raw session logs into a readable and shareable format.

Yes, this approach parses tool calls and session metadata from JSONL transcripts. It extracts user and assistant messages alongside tool interactions to ensure the resulting Markdown output accurately reflects the full session context.

You need DuckDB for indexing and searching Claude Code transcripts because it enables efficient cross-session analysis across large collections. This allows for quick querying and retrieval of specific conversations without manually opening individual files.