delegation-assessment
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill involves reading potentially untrusted implementation plans and repository metadata.
- Ingestion points: The skill reads Markdown and HTML plans, run manifests, and repository files.
- Boundary markers: No explicit delimiters are used to separate ingested data from instructions.
- Capability inventory: The skill is explicitly read-only and lacks capabilities for file modification, network access, or command execution.
- Sanitization: No sanitization is performed on ingested plan content.
- [SAFE]: The skill is designed as a read-only advisory tool. It employs security hardening through the disable-model-invocation frontmatter configuration, which limits its execution scope.
Audit Metadata