claw-history

What This Skill Does

The claw-history skill serves as the comprehensive auditing engine for an OpenClaw agent. Unlike standard session logs that only track current interactions, this skill is designed to perform a forensic reconstruction of the agent's entire operational lifespan. It bridges the gap between fragmented memory files, long-term milestones, and session logs to provide a transparent, linear narrative of the agent's journey from its inception ('birth') to the present moment.

Installation

To add this capability to your agent, execute the following command in your terminal or command-line interface within the agent environment:

clawhub install openclaw/skills/skills/bluebirdback/claw-history

Ensure that you have appropriate read permissions for the agent's memory directories, as this skill relies on accessing historical Markdown files and local logs to function effectively.

Use Cases

• Accountability & Auditing: Ideal for users who need to verify exactly when a specific task was initiated or completed over a long period.
• Agent Reflection: Helpful for debugging agent behavior by tracing back the root cause of certain operational habits to earlier sessions.
• Performance Review: Allows developers or users to see a timeline of when new skills were acquired, milestones were reached, and major changes occurred in the agent's configuration.
• Historical Context Retrieval: Useful when you forget the details of a previous project and need a chronological refresher on the steps taken months ago.

Example Prompts

1. "I need to review our progress. Can you provide a full chronological history of everything you've done since the day you were created?"
2. "Give me an account of my agent's lifetime timeline, highlighting the milestone markers from the very beginning until now."
3. "Please generate a complete action log covering my entire interaction history with you, including any sub-agent sessions you might have managed."

Tips & Limitations

• Data Dependency: The quality of the output is strictly dependent on the existence of memory/YYYY-MM-DD*.md files. If logs were cleared or not saved, the skill will identify these as 'gaps' in the final report.
• Precision: The skill distinguishes between 'Confirmed' entries backed by direct file evidence and 'Inferred' entries based on session order. Always check the confidence rating provided at the end of the report.
• Performance: For long-running agents with thousands of log entries, generating the history may take a moment. The skill is optimized to de-duplicate logs, but massive filesets can impact response latency. Use this skill when you specifically need the 'full view' rather than a quick lookup.