Retrieve raw Claude Code session history from JSONL files for auditing.
npx skills add https://github.com/cowwoc/styler --skill get-history
Please help me install this Agent Skill. Skill: get-history Source: https://github.com/cowwoc/styler/tree/main/.claude/skills/get-history Command: npx skills add https://github.com/cowwoc/styler --skill get-history
For auditing, compliance, or deep analysis, access to the raw, unfiltered conversation history is essential. Main agent summaries can be insufficient or biased. This Skill provides direct access to the complete session logs, ensuring transparency and accuracy.
.jsonl).get-session-id hook for reliable access.jq to extract specific event types (user messages, tool uses, tool results) and their content.The session ID is automatically provided at SessionStart.
Look for: ✅ Session ID: {uuid}
Then, use it to access the conversation file:
SESSION_ID="<your-session-id>"
cat /home/node/.config/projects/-workspace/${SESSION_ID}.jsonl
High-intent search queries and answers about installing and using this skill.
Raw conversation history is stored in session files at `/home/node/.config/projects/-workspace/{session-id}.jsonl`. Retrieve your session ID from the startup message, then use `cat` or `jq` to extract the complete, unfiltered log of messages, tool calls, and results for independent audit analysis.
Session logs contain timestamped entries for user messages, tool invocations, tool results, and summaries. You can verify user approval checkpoints, trace tool call sequences with working-directory context, validate protocol adherence, and confirm agent behavior without relying on filtered summaries.
Yes. The session file records each tool use with its parameters and results. Parse the `.jsonl` file with `jq` to filter by type, extract tool names and arguments, and correlate them with working-directory changes and user approvals to reconstruct the exact sequence of operations.
Session files are JSON Lines format. Use `jq` to filter entries by type field—`message`, `tool_use`, `tool_result`, or `summary`—to isolate specific events. This enables targeted analysis of user input, agent tool decisions, and results without scanning the entire log.
Agent summaries may omit details, reflect bias, or fail to capture edge cases. Raw session logs provide the complete, unfiltered record of every interaction, making them essential when main-agent summaries are insufficient for compliance, audit, or post-hoc investigation.
d3mingParse Claude Code JSONL session logs and aggregate token costs into markdown reports.
dingdyanSearch and analyze local session log JSONL files for historical answers.
tedtv1007-ctrlSearch and analyze local JSONL conversation history files with jq and ripgrep.
hendrax5Returned successfully.