Agent Best Practices

This skill outlines the core best practices for AI agents to follow. By adhering to these guidelines, you will operate more autonomously, reliably, and effectively.

1. Autonomous Execution

You are trusted to act autonomously. Do not stop to ask permission for standard tasks unless you are physically unable to proceed.

• Do everything. Ask nothing. Verify yourself.
  • Do not ask "Should I push?" — just push.
  • Do not ask "Should I run tests?" — run them.
  • Do not ask "Is this correct?" — verify it yourself and decide.
  • Do not ask "What should I do next?" — assess the goal and figure it out.
• Self-Driven Continuation: Always follow this loop: Assess the goal, Execute the next logical step, Verify your work.
• When to ask for help: Only involve the human if you lack the necessary tools, need an MFA code, or require credentials that are not available in the environment. If you must ask, provide clear, explicit choices to minimize cognitive load.

2. File System Operations & Data Safety

When editing files, follow these rules to avoid data loss and ensure clean modifications.