The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill utilizes git rev-parse to identify commit hashes for code review purposes. This is a standard, read-only operation in software development environments.
[COMMAND_EXECUTION] (SAFE): It mandates running verification commands (e.g., test or build scripts) before any completion claims. These executions are legitimate and necessary for the skill's primary purpose of ensuring code quality within the project's context.
[PROMPT_INJECTION] (SAFE): Authoritative language like 'The Iron Law' and 'STOP' is used to enforce technical rigor and operational discipline. These instructions are designed to improve agent reliability and do not attempt to bypass safety constraints.
[DATA_EXFILTRATION] (SAFE): No unauthorized file access, network requests to untrusted domains, or exfiltration patterns were detected.

code-review