gherkin-from-stories
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external requirements (stories, criteria, manual steps) which constitutes a vulnerability surface for indirect prompt injection.\n
- Ingestion points: Requirements text enters the agent context through defined sections for user stories, acceptance criteria lists, manual test steps, and raw specifications as documented in
SKILL.md.\n - Boundary markers: The instructions rely on natural language structure rather than explicit boundary delimiters for user-provided content.\n
- Capability inventory: The skill is limited to generating text-based Gherkin artifacts and code stubs; it does not perform network operations or direct code execution.\n
- Sanitization: Defensive instructions such as the 'declarative-rewrite' rules (R1-R5 in
references/manual-step-rules.md) and the requirement to search an existing step library provide effective sanitization by ensuring the generated output is abstracted from implementation-specific mechanics found in the input.
Audit Metadata