ln-51-acceptance-test-builder
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external requirements, repository instructions, and code files to design and implement tests, creating a potential surface for indirect injection.
- Ingestion points: Requirements documents, acceptance criteria, and existing codebase (Checklist Section 1).
- Boundary markers: The instructions require the agent to cite the exact source of instructions and resolve boundaries rather than assuming context (Section 1).
- Capability inventory: The skill utilizes shell commands, native code editors, and language servers for test implementation and execution (Tool Routing).
- Sanitization: The skill includes explicit instructions to inspect diffs for secrets, environment-specific paths, and unrelated changes before completion (Checklist Section 3).
- [COMMAND_EXECUTION]: The skill uses repository-defined shell commands for test execution and Git for workspace safety inspections.
- Evidence: These actions are strictly scoped to the primary purpose of the skill (acceptance testing) and include warnings against execution in production or unapproved environments (Tool Routing).
Audit Metadata