The Agent Skills Directory

[SAFE]: The skill serves as a configuration guide for a specific development environment and does not contain any executable instructions or automated scripts that would perform actions on the user's system.
[COMMAND_EXECUTION]: Provides example shell commands for building the project (npm install, npm run build) and testing authentication (op read, curl). These are standard developer operations and are presented as documentation for the user to copy/paste, not as automated triggers.
[CREDENTIALS_UNSAFE]: References the use of 1Password service tokens and auth.json files for credential management. It follows best practices by using secret management tools (op read) and local files instead of hardcoding API keys or environment variables. No actual secrets are exposed in the skill content.
[EXTERNAL_DOWNLOADS]: References the official Minimax API endpoint (api.minimax.io) and the upstream GitHub repository (badlogic/pi-mono). These are well-known services relevant to the skill's stated purpose.
[DYNAMIC_CONTEXT_INJECTION]: While the skill mentions the ! syntax used by the pi agent to execute commands for credential resolution (e.g., "key": "!op read ..."), it does not use the !command`` syntax within the SKILL.md file itself to execute code at load time.

pi-minimax-setup