sdd-feature
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-provided feature descriptions and project-level documentation to generate its output.
- Ingestion points: The 'feature_description' argument and the 'docs/project.md' file are used as primary context.
- Boundary markers: Absent; the skill does not wrap these external inputs in delimiters or provide 'ignore instructions' guards.
- Capability inventory: The skill's capabilities are limited to reading project context and writing markdown files (feature.md); it does not perform network operations or execute shell commands.
- Sanitization: Absent; there is no evidence of input validation or escaping before the data is processed by the model.
Audit Metadata