my
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides legitimate self-management functionality allowing the agent to optimize its performance. It includes constraints such as in-memory only storage and parameter validation for key settings.
- [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection. Ingestion points: Data is stored via the 'my' tool's 'set' action in SKILL.md. Boundary markers: Absent for values stored in the scratchpad. Capability inventory: The 'my' tool can modify the 'model', 'max_iterations', and 'context_window_tokens' as defined in SKILL.md. Sanitization: SKILL.md specifies that protected parameters have type and range validation. The instructions mandate informing the user before making changes, which acts as a safeguard against malicious reconfiguration.
Audit Metadata