gogo2/tools.md

Available Tools

This document lists all the tools available in the AI coding assistant.

File Operations

• read_file: Read contents of a file from the local filesystem
• write: Write a file to the local filesystem (overwrites existing files)
• search_replace: Perform exact string replacements in files
• MultiEdit: Make multiple edits to a single file in one operation
• delete_file: Delete a file at the specified path
• list_dir: List files and directories in a given path
• glob_file_search: Search for files matching a glob pattern

Code Analysis and Search

• codebase_search: Semantic search tool to find code snippets relevant to queries
• grep: Powerful search tool built on ripgrep for exact symbol/string searches
• read_lints: Read and display linter errors from the current workspace

Terminal Operations

• run_terminal_cmd: Execute terminal commands on behalf of the user

Web and External Resources

• web_search: Search the web for real-time information about any topic
• fetch_rules: Fetch rules provided by the user to help with navigating the codebase

Task Management

• todo_write: Create and manage a structured task list for coding sessions

Notebook Operations

• edit_notebook: Edit Jupyter notebook cells (supports editing existing cells and creating new ones)

Git Operations

• fetch_pull_request: Fetch pull request information including title, description, and diff

Memory Management

• update_memory: Create, update, or delete memories in a persistent knowledge base

Diagram Creation

• create_diagram: Create Mermaid diagrams

Tool Usage Guidelines

File Operations

• Always prefer editing existing files rather than creating new ones
• Use MultiEdit for multiple changes to the same file
• Be careful with file paths - use relative paths when possible

Code Search

• Use codebase_search for semantic queries about functionality
• Use grep for exact string/symbol searches
• Check linter errors before and after making changes

Terminal Commands

• Commands run in the same shell session unless specified otherwise
• Use non-interactive flags for commands that require user input
• Background long-running processes appropriately

Task Management

• Use todo_write for complex multi-step tasks
• Update task status in real-time
• Only one task should be in_progress at a time

Memory Management

• Use update_memory for persistent information that should be remembered across sessions
• Include relevant context and metadata
• Use appropriate action types (create, update, delete)

Web Search

• Use for real-time information not available in training data
• Include version numbers or dates for technical queries
• Verify current facts and updates

Notebook Editing

• Use edit_notebook exclusively for Jupyter notebook operations
• Provide extensive context for unique identification
• Support both editing existing cells and creating new ones

Diagram Creation

• Use create_diagram for Mermaid diagrams
• Specify diagram type and content clearly
• Diagrams are rendered as SVG images

Git Operations

• Use fetch_pull_request to get detailed PR information
• Includes title, description, diff, and commit details
• Useful for understanding code changes and context

Best Practices

• Batch tool calls when possible for better performance
• Always read files before editing them
• Use appropriate tools for the task at hand
• Follow the user's coding rules and preferences
• Maintain code quality and avoid creating duplicate implementations