self-eval
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes the
$ARGUMENTSvariable and the full conversation history to generate evaluations, creating a surface where malicious instructions embedded in the analyzed content could be executed by the agent. - Ingestion points: User-supplied arguments via the
$ARGUMENTSplaceholder and the conversation history (SKILL.md). - Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands in the data it evaluates.
- Capability inventory: The skill has file system access to read and write the
.self-eval-scores.jsonlfile in the working directory. - Sanitization: No sanitization or filtering logic is present to handle potentially malicious input within the evaluation data.
Audit Metadata