error-debugging-multi-agent-review
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external software artifacts (code snippets, files, and repositories) provided through the
$ARGUMENTSparameter. This creates a surface for indirect prompt injection attacks. - Ingestion points: The
$ARGUMENTSparameter inSKILL.mdis used to pass target code for review. - Boundary markers: The instructions do not specify the use of clear delimiters or explicit "ignore embedded instructions" directives to help the agent distinguish between the code it is reviewing and instructions that might be hidden within that code.
- Capability inventory: The skill orchestrates multiple specialized agents (Security, Architecture, Performance, etc.), which involves complex reasoning and synthesis of results, increasing the potential impact if an agent is misled by malicious code comments.
- Sanitization: There is no mention of sanitizing or filtering the input code to remove potential injection vectors before processing.
Audit Metadata