Heleni Self Learning

This skill is intended to log events and turn recurring mistakes into persistent rules by editing files under ~/.openclaw/workspace (LEARNINGS.md, SOUL.md, etc.). Before installing or enabling it: - Understand persistence: the skill's scripts append rules to SOUL.md, so learnings can permanently change agent behaviour. Decide whether you want automatic promotion or manual review. - Protect secrets: the skill explicitly says to promote 'Contact / credentials' to MEMORY.md. Do NOT store passwords, API keys, or other secrets in plaintext files. Use a secure secret store instead, or remove that advice. - Review file writes: inspect or run the provided scripts in a sandbox to confirm they do only what you expect. Keep backups of SOUL.md and other config files so you can revert unwanted changes. - Autonomy controls: if the agent can invoke skills autonomously, consider restricting this skill from auto-promoting rules or require owner approval for changes to SOUL.md. - Compatibility note: some commands (e.g., grep -P, GNU sed flags) may not behave the same on all systems; test on your environment. If you want a lower-risk setup: use the logging portion (write logs) but require manual promotion into SOUL.md, and never store credentials in workspace files. If you can provide more metadata (who published this skill, a homepage, or an explicit declaration of the config paths it will modify), I can raise or lower my confidence.