error-debugging-multi-agent-review
Pass
Audited by Gen Agent Trust Hub on Aug 9, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill handles untrusted external code through the
$ARGUMENTSparameter without defining boundary markers or sanitization protocols, which allows for potential indirect prompt injection.\n- Ingestion points: The project code or Git repositories are ingested through the$ARGUMENTSvariable defined inSKILL.md.\n- Boundary markers: The skill does not provide instructions to wrap the ingested code in delimiters or to ignore embedded natural language instructions.\n- Capability inventory: The agent context for this skill involves reading and analyzing file systems and repository contents.\n- Sanitization: There is no instruction to escape, filter, or validate the content of the ingested code before analysis by the agent network.
Audit Metadata