The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided environment and target details to generate CI configuration files. It mitigates potential injection risks by instructing the agent to use the provided references/schema.json to validate all generated configurations using the mirrord verify-config command before presenting them to the user.
[COMMAND_EXECUTION]: The skill provides numerous examples of shell commands for CI workflows, including mirrord ci start/stop and kubectl. It maintains a strong security posture by explicitly warning users against using insecure installation patterns like piping remote scripts to a shell (curl | bash) and recommending the use of trusted CI images or pinned versions instead.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill handles sensitive CI components like KUBECONFIG and MIRRORD_CI_API_KEY by correctly instructing the user to store them as secrets within their CI platform's environment, rather than hardcoding them.

mirrord-ci