inner-life-reflect

What This Skill Does

inner-life-reflect introduces a sophisticated meta-cognitive layer to your OpenClaw agent, moving it from a static execution model to a learning system. Most agents operate on a blank slate with every interaction, repeating identical reasoning errors and ignoring subtle preference shifts. This skill implements a structured reflection pipeline using trigger detection and a rigorous four-point quality gate. By observing its own behavior, the agent logs meaningful growth into SELF.md only when specific thresholds are met. It distinguishes between foundational identity (SOUL.md) and evolving patterns (SELF.md), ensuring your agent matures over time rather than accumulating noise.

Installation

Before proceeding, ensure you have the core foundation installed. Execute the following command: clawhub install openclaw/skills/skills/dkistenev/inner-life-reflect

Prerequisites: You must have inner-life-core initialized first. Verify the presence of memory/inner-state.json and memory/habits.json. If missing, run: clawhub install inner-life-core followed by bash skills/inner-life-core/scripts/init.sh.

Use Cases

• Continuous Personalization: The agent learns your specific communication cadence and preferences without you explicitly repeating them.
• Bias Mitigation: By flagging repeated avoidance patterns or incorrect reasoning styles, the agent can self-correct its logic in future sessions.
• Long-term Project Continuity: The agent tracks how its approach to technical tasks evolves, ensuring it doesn't revert to suboptimal methods used in early project phases.

Example Prompts

1. "Reflect on our interactions this week—have you noticed a pattern in how you handle my code requests?"
2. "Perform a micro-check of our recent session; is there any behavior that warrants an entry in SELF.md?"
3. "Review the last month of my instructions and tell me how your operational persona has shifted."

Tips & Limitations

• Trust the Gate: Do not force the agent to write entries. The quality gate (Specificity, Evidence, Novelty, Usefulness) is intentional. It prevents the agent from filling your memory with generic platitudes.
• Monitor the Evolution: Periodically review the Evolution section in SELF.md to ensure the agent’s trajectory aligns with your expectations.
• Manual Pruning: While the skill manages its own logs, if you notice the agent reflecting on irrelevant data, you can manually adjust the trigger sensitivity in your configuration file.