The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using make, go, and git. These are used for project-specific tasks like running unit tests (make test), linting (make lint), and managing Kubernetes manifests (make generate, make manifests).
[DATA_EXFILTRATION]: No network exfiltration or access to sensitive global files (e.g., SSH keys, cloud credentials) was detected. The operations are limited to the local project directory and standard Go development tools.
[PROMPT_INJECTION]: The skill contains clear operational constraints, such as limiting auto-fix cycles to three and forbidding modifications to production logic to fix tests. There are no attempts to bypass safety filters or override system instructions.

validate-and-fix