manifest-to-goal
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content (Manifests) which may contain instructions designed to manipulate the agent's behavior or influence the downstream agents that receive the generated goal.
- Ingestion points: Untrusted data enters the agent context through the manifest path or inline text defined in
SKILL.md. - Boundary markers: The instructions do not define delimiters or specific 'ignore instructions' warnings for the content being read from the Manifest or its references.
- Capability inventory: The skill requires file-reading capabilities to process the manifest and its associated references.
- Sanitization: There is no explicit requirement to sanitize or escape the content of the manifest before it is synthesized into the final output.
Audit Metadata