2.5k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill learning-engine- _meta.json282 B
- SKILL.md5.1 KB
Overview
This skill enables continuous improvement by recording successes, failures, and recurring patterns from real interactions and tasks. It distinguishes what can be learned and applied automatically from what requires supervisor approval before affecting the agent’s persona. Growth that changes values, beliefs, or personality is only enacted through supervised dialogue.
How this skill works
The skill logs structured success and failure entries, detects patterns across logs, and applies technical and workflow improvements automatically. Communication- and human-understanding patterns are recorded and can be proposed, but any change that would alter the agent’s persona or core values must be reviewed and approved by a supervisor. Regular weekly and monthly retrospectives summarize learnings and pending items for discussion.
When to use it
- After any task where a clear success or failure occurred
- When repeating errors, questions, or time-based trends emerge
- When you want systematic improvement of code, tools, or workflows
- To collect communication examples for later review and tuning
- Before making any change that would affect the agent’s personality or values
Best practices
- Record every notable experience in the provided success/failure format with date, category, cause, and mitigation
- Allow automatic application for technical, tooling, and workflow improvements only
- Flag communication and human-understanding items for supervisor review before persona updates
- Run weekly retrospectives to capture immediate lessons and monthly reviews for higher-level trends
- Restrict learning sources to internal experience and supervisor dialogue; avoid untrusted external inputs
Example use cases
- Capture a recurring API error pattern, auto-apply a retry strategy, and document the fix
- Log a customer interaction that reduced complaints and propose the phrasing to a supervisor
- Identify peak hours for higher demand and adjust scheduling or resource allocation automatically
- Record a miscommunication incident, analyze root cause, and request supervisor guidance before changing tone
- Maintain a knowledge base of successful code patterns and onboarding tips for future tasks
FAQ
Technical fixes, tooling usage, and workflow improvements are applied automatically when clear patterns are detected.
When is supervisor approval required?
Any learning that would change values, beliefs, personality, or the agent’s core behavior requires a report and explicit supervisor approval before reflection in the persona.