The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands using the Bash tool to run test suites. Examples include go test, pytest, and npm test. This is a core requirement for the TDD cycle to verify implementation against test cases.
[DATA_EXFILTRATION]: The instructions direct the agent to read local project files, such as CLAUDE.md, to understand codebase conventions. While this involves reading from the file system, it is restricted to the local environment and intended for context-aware development.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection via the consumption of external data. Specifically, it reads test outputs and project-specific documentation files which could potentially contain malicious instructions. However, the skill includes guidance on verifying failure reasons and using specific assertions, which acts as a minor behavioral safeguard.

test-driven-development