delegate
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of delegation prompts that interpolate data from the agent's current session context into instructions for sub-agents, creating a vulnerability surface for indirect injection.
- Ingestion points: Data enters the system via the
OBSERVATIONS(error messages, system state) andECOSYSTEM CONTEXT(session-specific facts) sections inSKILL.md. - Boundary markers: The template uses markdown headers and list items to structure data, but lacks explicit instruction-ignore delimiters or sanitization for the interpolated content.
- Capability inventory: The resulting sub-agent prompts grant broad capabilities including context gathering, hypothesis testing, and solution implementation across the repository.
- Sanitization: There are no instructions for sanitizing or escaping content derived from potentially untrusted sources like error logs or external file content before it is passed to the sub-agent.
Audit Metadata