What problem does it solve?
Recover and inspect historical conversation context that is not present in active memory by searching append-only session JSONL files, enabling continuity across parent/older chats and audit of past interactions.
Core Features & Use Cases
- Indexed session discovery: Locate sessions by date, provider, or filename using the sessions.json index and file timestamps.
- Message extraction and filtering: Extract user or assistant text content, tool calls, and compute message counts or token metrics from JSONL records.
- Cost and usage analysis: Aggregate per-session cost, produce daily summaries, and generate tool-usage breakdowns for billing or optimization.
- Use Case: Investigate what a user asked in a previous chat, calculate the cost of a long-running session, or search across all agent sessions for a specific phrase.
Quick Start
Search the agent's session directory for the phrase billing and summarize matching conversations with timestamps and total cost.