The Agent Skills Directory

[SAFE]: The skill instructions describe a legitimate software development process focused on behavior verification through public interfaces. No malicious patterns, prompt injections, or obfuscated content were detected.
[COMMAND_EXECUTION]: The skill utilizes shell access (Bash) for the intended purpose of running tests and committing code changes, which is consistent with its stated utility as a development tool.
[INDIRECT_PROMPT_INJECTION]: While the skill interacts with untrusted data (existing source code), it follows a standard development workflow with human-in-the-loop checkpoints.
Ingestion points: Reads files and directory structures using Read, Grep, and Glob tools.
Boundary markers: None explicitly defined for input data, relying on standard development context.
Capability inventory: Full shell access (Bash), file creation (Write), and file modification (Edit).
Sanitization: Not applicable as the skill operates on local source code as part of a development task.

tdd