test-driven-development
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: Behavioral Override. The skill uses absolute directives such as "The Iron Law," "No exceptions," and "Delete means delete" to enforce its methodology. It explicitly commands the agent to delete user-written code if it fails to follow the TDD sequence, which overrides typical AI safety behavior regarding the preservation of user data.
- [PROMPT_INJECTION]: Indirect Injection Surface. The skill requires the agent to ingest and evaluate the content of user-provided source files and test scripts to verify process adherence.
- Ingestion points: Processes local project files (e.g.,
path/to/test.test.ts) during verification steps. - Boundary markers: The instructions do not define delimiters or "ignore instructions" wrappers to separate untrusted file content from the agent's logic.
- Capability inventory: The agent context involves file-system access (read/write/delete) and shell command execution.
- Sanitization: No validation or sanitization of the content of the analyzed files is present.
- [COMMAND_EXECUTION]: Development Tooling. The skill provides instructions for executing standard development commands, specifically
npm test, to facilitate the TDD cycle. While these are legitimate tools, they are part of the execution flow that evaluates user-provided code.
Audit Metadata