Skills
skills/openclaw/skills/self-improvement/Gen Agent Trust Hub

self-improvement

Pass

Audited by Gen Agent Trust Hub on Mar 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill implements a behavioral feedback loop by injecting specific instructions and reminders into the agent's session context.
  • The activator.sh script and OpenClaw handlers (handler.js, handler.ts) are designed to insert "Self-Improvement Reminders" into the prompt stream.
  • Instructions guide the agent to self-modify its behavioral guidelines and operational workflows based on session discoveries.
  • [COMMAND_EXECUTION]: Local shell scripts are provided to automate maintenance and monitor tool outputs.
  • error-detector.sh scans CLAUDE_TOOL_OUTPUT for failure patterns to trigger logging reminders.
  • extract-skill.sh creates directories and templates for new skills based on user-provided names.
  • Documentation facilitates the setup of these scripts as automated hooks within the agent's runtime environment.
  • [EXTERNAL_DOWNLOADS]: Setup instructions reference external source repositories for installation.
  • The documentation suggests fetching the skill via git clone from a GitHub repository (github.com/peterskoett/self-improving-agent.git).
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 17, 2026, 08:27 AM