lena-learning

What This Skill Does

The lena-learning skill transforms your OpenClaw agent from a static tool into an evolving intelligence partner. By implementing a sophisticated feedback loop and a structured memory architecture, the agent captures every interaction as an opportunity to refine its performance. It automatically parses conversations to extract key insights, documents errors to prevent repetition, and catalogs user preferences to ensure future responses are tailored specifically to your needs.

At its core, the skill manages a multi-tier memory system. Daily activity is recorded in dated log files, while long-term knowledge, personal preferences, and tool-related environment settings are archived in specific documentation files (MEMORY.md, USER.md, TOOLS.md). This structured approach allows the agent to build a persistent profile of its environment, meaning it learns not just what you want, but how you prefer your workflows handled.

Installation

To integrate this intelligence layer into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/bwtomekk-bit/lena-learning

Ensure that your environment allows file-system operations, as this skill relies on read/write access to maintain its memory logs.

Use Cases

1. Personalization Engine: Automatically remember communication styles, tone preferences, and formatting requirements so you don't have to repeat them.
2. Error Mitigation: Document technical errors or logical failures in a persistent file, allowing the agent to analyze previous failure states and adjust its logic accordingly.
3. Knowledge Retention: Keep track of complex project details or recurring instructions without manual configuration updates.
4. Workflow Optimization: Identify patterns in your requests to suggest improvements to tool usage or task sequencing.

Example Prompts

1. "Lena, remember that I always prefer my project summaries in bullet points with a brief executive summary at the top."
2. "Lena, we tried this technical approach yesterday and it failed with a timeout; make sure to document this in TOOLS.md so we don't try it again."
3. "Review our session from earlier today and identify three areas where my instructions could have been clearer to improve your efficiency."

Tips & Limitations

• Transparency: Regularly review the memory/ directory. Since the agent writes its own logs, periodic inspection ensures the long-term memory remains clean and relevant.
• Feedback is Key: The agent learns best when you explicitly point out corrections. Treat every 'correction' as a training signal.
• Storage Limits: Be mindful of file sizes if you interact with the agent thousands of times daily; the system is designed for intelligence, not high-volume data storage.
• Contextual Awareness: While the skill is powerful, it works best when the user is intentional about sharing feedback. The more you teach it, the more autonomous it becomes.