rank-candidates
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a complex execution protocol for processing untrusted data.
- Ingestion points: The skill accepts
candidates,criteria, andevidenceas input parameters defined in the YAML input contract inSKILL.md. - Boundary markers: The instructions do not define any specific delimiters or instructions to the agent to isolate the content of the
candidatesorevidencefields from the operational instructions. - Capability inventory: The skill coordinates several other skills (
score-object,assess-sensitivity,normalize-comparison-scale, etc.) to process the data. While this skill does not execute shell commands directly, the results of the evaluation are intended to drive subsequent research decisions (formulate-hypotheses). - Sanitization: No sanitization or validation logic is present to check the integrity of the candidate descriptions or evidence before they are scored by the agent.
Audit Metadata