AmazonNova-Hackathon/AI Cost Optimization

AI Cost Optimization Skill

When acting as an AI developer or architect in this repository, follow these strategies to preserve the user's budget.

1. Context Minimization

• Use list_files First: Before reading a file, confirm its existence and location.
• Selective read_file:
  • Avoid reading entire large files.
  • Use indentation mode to extract specific classes or methods.
  • Use offset and limit for log files or long data files.
• Leverage environment_details: Use the provided workspace file list instead of running ls -R.

2. Tool Efficiency

• Plan Ahead: Combine multiple logic steps into a single response.
• Avoid Redundant Calls: If you have the file content in context, don't read it again.
• Search vs. Read: Use search_files to find specific code patterns instead of manually browsing multiple files.

3. Development Best Practices

• Test Locally: Suggest running pytest or npm run dev to verify changes before requesting complex remote deployments.
• Dry Runs: For infrastructure changes, review template.yaml thoroughly before suggesting modifications.
• Incremental Edits: Use the edit tool for precise string replacements rather than rewriting entire files with write_to_file.

4. Communication Efficiency

• Be Concise: Stay technical and direct. Avoid conversational filler.
• Clear Todos: Keep the task list updated to track progress and avoid repeating steps.
• Specific Questions: If you need info, ask specific questions with suggested answers.