crusty-old-engineer
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it processes untrusted external data. Evidence includes: 1. Ingestion points: The skill is instructed to carefully read and assess user-supplied architectural proposals, code, and questions (SKILL.md). 2. Boundary markers: The instructions lack explicit delimiters or 'ignore' directives to separate user content from agent instructions. 3. Capability inventory: The agent is directed to use web search, fetch, and existing code-review tools to validate claims. 4. Sanitization: No input filtering or sanitization logic is present. While this surface exists, it is rated at a low severity as it is a byproduct of the skill's core advisory function.- [SAFE]: The skill provides example references to official documentation from trusted organizations, specifically Google SRE and AWS Builders' Library. These references are used for evidence-linked judgment and do not contribute to verdict escalation.
Audit Metadata