Precision Mastery

What it does

Precision Mastery is a skill designed to optimize token usage when an AI agent performs file operations like reading, editing, writing, searching (grep), and web fetching. It provides alternatives to standard tool behaviors that significantly reduce token consumption—saving 75-95% of tokens on these tasks. The skill teaches best practices for using these optimized methods effectively.

When to use it

• When performing file operations: If your agent needs to read, edit, write, or search files.
• To reduce token costs: Especially when dealing with large files or frequent operations.
• For code analysis and refactoring: To understand file structure and identify patterns without needing the full content of each file.
• When building import statements: To quickly find exported symbols from other files.

Key capabilities

• Token-efficient alternatives to standard tools: Offers count_only, minimal, files_only, and paths_only verbosity options for various operations (write, edit, read, grep, glob, fetch).
• Extract Modes for Reading: Provides modes like content, outline, symbols, ast, and lines to selectively extract information from files.
• Verbosity Cheat Sheet: Provides guidance on selecting the lowest verbosity level needed for each operation.
• Token Multipliers: Defines token usage multipliers associated with different verbosity levels (count_only, minimal, standard, verbose).

Example prompts

• "Use Precision Mastery to read src/components/Button.tsx in outline mode."
• "With Precision Mastery, grep for the word 'error' within files only and output just the file paths."
• "Edit config.json using Precision Mastery with minimal verbosity."

Tips & gotchas

• Start with Outline: When analyzing a file, begin by reading it in outline mode to understand its structure before potentially needing full content.
• Understand Verbosity Levels: Lower verbosity levels (like count_only) significantly reduce token usage but might require you to skip confirmation steps.
• Use Extract Modes strategically: Choose the appropriate extract mode (content, outline, symbols, etc.) based on what information is needed from a file.