combinatorial-testing
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions are strictly limited to static analysis and design tasks. The manifest and prompt files include explicit prohibitions against executing code, accessing external networks, or modifying the target environment, ensuring the agent remains within a safe boundary (SKILL.md, prompts/combinatorial-testing.md).
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied testing material but includes robust protections against adversarial input. 1. Ingestion points: Material enters via the user-supplied factors and constraints (prompts/combinatorial-testing.md). 2. Boundary markers: The skill enforces a 'six-part input audit' to explicitly categorize provided data and identify gaps or assumptions (SKILL.md). 3. Capability inventory: The skill contains no instructions for subprocess calls, file writes, or network operations, and explicitly forbids them (SKILL.md). 4. Sanitization: The required output format forces a separation of facts from inferences and recommendations, preventing the agent from interpreting data as new instructions (prompts/combinatorial-testing.md).
Audit Metadata