test-driven-development
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process code and requirements provided by the user, creating an inherent surface for indirect prompt injection.\n
- Ingestion points: The agent reads project files and writes implementation code based on user-defined tasks as guided by SKILL.md.\n
- Boundary markers: The workflow utilizes phase gates (RED, GREEN, REFACTOR) to maintain cycle discipline, but does not specify explicit prompt delimiters for isolating untrusted data.\n
- Capability inventory: The skill requires high-privilege tools such as Bash for test execution and Write for file manipulation, which are appropriate for its development purpose.\n
- Sanitization: There are no instructions for sanitizing or escaping the user-provided code strings before they are written to disk or executed in the shell.\n- [DYNAMIC_EXECUTION]: The local helper script scripts/validate.py uses the Python compile() function.\n
- Evidence: The function validate_script_syntax reads the content of the validator script itself and passes it to compile() to verify the code's syntax.\n
- Context: This dynamic execution is strictly limited to the script's own source code for validation purposes and does not process external payloads or execute the resulting object, posing no security risk.
Audit Metadata