alignment
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection by processing external data from project files that may contain malicious instructions intended to influence the agent's behavior or document synthesis.
- Ingestion points: Content is read from
design-docs/NN-name.brief.md,design-docs/NN-name.research.md, anddocs/glossary.mdduring Phase 1. - Boundary markers: There are no explicit instructions to use delimiters or "ignore" directives when interpolating content from these files into the synthesis prompt.
- Capability inventory: The skill possesses file-write capabilities to the local file system, specifically creating and updating documentation files in the
design-docs/anddocs/directories. - Sanitization: The process does not define any validation, escaping, or filtering mechanisms for the data retrieved from external documents before it is used to generate the final alignment artifacts.
Audit Metadata