Session

Search and analyze Claude Code conversation history via a DuckDB index over JSONL session files.

Current Session ID: ${CLAUDE_SESSION_ID}

Running Queries

The index refreshes automatically on first use per session. Subsequent queries skip the refresh for faster results. Pass --refresh to force a re-scan when the user asks for the latest data.

${CLAUDE_SKILL_DIR}/scripts/query.ts "SELECT model, SUM(output_tokens) as tokens FROM messages WHERE type = 'assistant' GROUP BY model"
${CLAUDE_SKILL_DIR}/scripts/query.ts --refresh "SELECT * FROM sessions ORDER BY start_time DESC LIMIT 5"

Named Queries

Built-in queries in resources/queries/ can be run by name with key=value params. Prefer these over writing SQL from scratch for common tasks: