agent-self-evaluation
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content, including agent outputs and task descriptions, to generate evaluation reports and instructions for improvement.
- Ingestion points: The
scripts/evaluate.pyscript reads data from stdin or user-specified files via the--taskand--outputarguments. - Boundary markers: There are no explicit delimiters or protective instructions used to separate untrusted data from the scoring logic.
- Capability inventory: The skill performs text analysis using regular expressions and prints results; it does not have network access or capabilities to execute the analyzed content as code.
- Sanitization: The tool does not sanitize or escape the external content it evaluates.
Audit Metadata