test-driven-development
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external source code and repository data, creating an attack surface where untrusted data enters the agent context.
- Ingestion points: Ingests user-provided source code files and test suites during the development workflow.
- Boundary markers: The instructions lack specific delimiters or boundary markers to differentiate between executable code and potentially malicious instructions embedded in comments or data.
- Capability inventory: The agent has permissions to write, modify, and delete files within the project scope (e.g., executing the Red-Green-Refactor cycle).
- Sanitization: There is no explicit sanitization or validation logic applied to the content of the files being processed.
- [COMMAND_EXECUTION]: The skill contains a directive to delete production code written in the current session if it was not first verified by a failing test. While this is a functional requirement of the TDD methodology, it involves file deletion capabilities. This is mitigated by explicit instructions to preserve all pre-existing or user-authored code and to seek user confirmation if the code's origin is unclear.
Audit Metadata