conversation-reader

To extract user and assistant messages from Claude JSONL logs, you provide the file path and apply composable filters to selectively isolate specific message streams, ensuring deterministic output for verification.

Auditing past Claude conversations across multiple projects is best handled by applying composable filters to JSONL logs, selectively extracting user messages, assistant responses, and tool calls for compliance review.

Yes, you can filter JSONL conversation history to only show tool calls by applying the specific composable filters designed to include or exclude individual message types like tool usage.

Extracting conversation history with this approach works efficiently with large log files through deterministic output generation and optimized handling of large JSONL session histories.

You should use composable filters instead of manually reading JSONL session logs because manual extraction is tedious and error-prone, whereas filters guarantee deterministic output and accurate message isolation.

To verify tool usage and assistant responses in Claude session logs, you extract the desired message streams from your JSONL file using composable filters to isolate specific data for review.