The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is instructed to run the project's test suite and linting tools both before and after modifications. This is standard behavior for development-focused agents to ensure code quality and functional parity.
[DATA_EXPOSURE]: The skill accesses local files (e.g., AGENTS.md, CONTRIBUTING.md, and source code) to understand project conventions. These operations are restricted to the local repository environment and no network exfiltration patterns were detected.
[SAFE]: No obfuscation, prompt injection, or remote code execution vulnerabilities were identified. The instructions specifically include safety guardrails, such as prohibiting automatic commits and stopping if the diff exceeds a reasonable size.

code-simplifier