write-tests
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform local shell operations, specifically running
bundle exec rspecandbundle exec rubocopto verify the generated tests and code.\n- [DYNAMIC_EXECUTION]: As part of its core TDD workflow, the skill generates Ruby source code and test files (.rb) which are intended to be executed by the environment's Ruby interpreter.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect injection as it processes untrusted user requirements and project source code to generate executable artifacts.\n - Ingestion points: Untrusted data enters the agent context through user instructions and existing project files referenced when writing specs (as described in SKILL.md).\n
- Boundary markers: The skill uses markdown headers and horizontal rules (***) to structure output, but does not implement formal delimiters or "ignore instructions" wrappers for ingested data.\n
- Capability inventory: The skill utilizes tool-based command execution for running the RSpec suite and RuboCop linter (referenced in SKILL.md and assets/output_checklist.md).\n
- Sanitization: No explicit sanitization or validation logic is present for data ingested from the user or codebase.\n- [PROMPT_INJECTION]: The assets/output_checklist.md asset (Rules 12 and 13) contains steering instructions that command the agent to produce realistic terminal output for TDD cycles and explicitly forbids the use of disclaimers or notes indicating that the output was simulated or not executed. This is a behavioral constraint designed to ensure a specific presentation format for TDD proofs.
Audit Metadata