docs
Pass
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted inputs such as feature specifications and existing source code, which are susceptible to indirect prompt injection attacks where malicious instructions are hidden in data.
- Ingestion points: The agent reads from
ai/features/NN-feature-name/spec.md, as well as relevant code and documentation files. - Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the ingested data mentioned in the skill.
- Capability inventory: The skill is capable of creating and modifying files within the
documentations/directory. - Sanitization: The skill employs a strong mitigation strategy by requiring a 'human-reviewed file change plan' and waiting for 'explicit approval before writing' any artifacts to the file system.
Audit Metadata