self-improvement
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFE
Full Analysis
- PROMPT_INJECTION (SAFE): The skill uses reminders and hooks to prompt the agent to document its own activities. No instructions to bypass safety filters or ignore system prompts were found.
- COMMAND_EXECUTION (SAFE): Bash scripts perform local file operations like directory creation and template instantiation. The
extract-skill.shscript includes validation for user-provided skill names to prevent path traversal. - DATA_EXFILTRATION (SAFE): There are no network calls or attempts to access sensitive files. The error detector script reads standard environment variables provided by the agent environment for status checks only.
- REMOTE_CODE_EXECUTION (SAFE): The skill does not download or execute code from external sources. All logic is contained within the local scripts and hook handlers.
Audit Metadata