session-logs

To search conversation history for a specific keyword, this Skill uses the rg tool to scan stored session JSONL logs and extract matching lines. It filters by message role to help you quickly recover prior context that is missing from your current memory.

Yes, you can compute token counts and cost analysis from JSONL session logs. The Skill uses jq to parse and aggregate JSONL records, summarizing per-session total cost and extracting tool usage breakdowns for comprehensive cost analysis.

Yes, you need jq and rg installed to search and filter chat history. These dependencies are required to deterministically parse, filter, and aggregate the JSONL records stored in your local session directory.

The best way to list previous sessions and extract human-readable text is by using this Skill to locate relevant JSONL files under your agent's sessions directory. It parses the records with jq to extract specific message roles and content.

Searching session logs requires local access to the agents directory because the conversation transcripts are stored as JSONL files under ~/.clawdbot/agents/<agentId>/sessions. The Skill reads these local files directly to ensure deterministic parsing and aggregation.