The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions specify running the project's full test suite and type-checking utilities (such as TypeScript, mypy, or pyright) locally. This involves executing code within the local environment to verify branch readiness.
[PROMPT_INJECTION]: The skill processes branch diffs and commit messages, which are external ingestion points. It lacks explicit boundary markers or sanitization for this content. The capabilities involved include git commands and test execution. This presents a surface for indirect prompt injection where malicious diff content could attempt to influence the audit outcome.
[SAFE]: The skill includes a specific dimension to scan commit messages for secrets and credentials, providing a security check against accidental data exposure in the repository history.

synthesis-preflight