specialist-engineering-test-writer-fixer
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The sub-agent is instructed to execute shell commands using various test runners such as Jest, Pytest, Mocha, and PHPUnit to identify affected files and verify code changes.\n- [PROMPT_INJECTION]: The skill interpolates the raw user input
{the user's question}directly into the prompt for the sub-agent. This lack of boundary markers (e.g., XML tags or clear delimiters) creates a surface for indirect prompt injection where a user could potentially supply instructions that override the agent's defined behavior.\n - Ingestion points: Interpolation into the sub-agent prompt in
SKILL.md.\n - Boundary markers: Absent.\n
- Capability inventory: File system modification (writing tests) and shell command execution (running tests).\n
- Sanitization: None observed in the static definition.
Audit Metadata