verify-security-finding
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external security findings which creates an attack surface for indirect prompt injection.
- Ingestion points: The agent is directed to pass every material candidate from external sources through the verification skill.
- Boundary markers: No delimiters or boundary instructions are defined to separate the untrusted findings from the agent instructions.
- Capability inventory: The agent is tasked with reproduction, search for variants, and updating project files (plan.md), which involves active tool usage.
- Sanitization: No sanitization or validation of the input candidates is required before processing.
- [NO_CODE]: The skill consists entirely of markdown-based documentation and metadata, containing no executable code or external script references.
Audit Metadata