self-improvement
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill includes shell scripts (scripts/activator.sh, scripts/error-detector.sh, scripts/extract-skill.sh) intended for local execution by the agent. scripts/extract-skill.sh performs directory and file creation based on a defined template. It implements security checks including regex validation for skill names and path sanitization to prevent directory traversal or writes outside the project directory.
- [PROMPT_INJECTION]: The skill implements a self-improvement workflow that ingests untrusted data from tool outputs (via CLAUDE_TOOL_OUTPUT) and user interactions into log files (e.g., .learnings/LEARNINGS.md) which are later read as context. Boundary markers are present in the form of markdown headers and structured metadata fields to help delineate ingested content. The capability inventory includes the ability to write files and execute local scripts, but these operations are restricted by the aforementioned sanitization logic.
Audit Metadata