Home / Tools

🛠️ Tools Reference

Code Puppy agents use tools to interact with your system. Each agent has access to a specific set of tools based on its purpose.

💡 Tip

Use /tools to see what tools are available to the current agent.

📁 File Operations

Tools for reading, writing, and managing files.

list_files(directory=".", recursive=True)

List files and directories with intelligent filtering. Automatically ignores common build artifacts and cache directories.

Parameters:

directory - Path to list (default: current directory)
recursive - Whether to list subdirectories (default: True)

read_file(file_path, start_line=None, num_lines=None)

Read file contents with optional line-range selection for handling large files.

Parameters:

file_path - Path to the file to read
start_line - Starting line number (1-based, optional)
num_lines - Number of lines to read (optional)

edit_file(payload)

Comprehensive file editing supporting multiple strategies:

ContentPayload - Create or overwrite a file with content
ReplacementsPayload - Targeted text replacements
DeleteSnippetPayload - Remove specific text snippets

delete_file(file_path)

Safely delete a file with automatic diff generation showing what was removed.

grep(search_string, directory=".")

Recursively search for text patterns using ripgrep. Supports regex and various flags.

Parameters:

search_string - Pattern to search for (supports rg flags)
directory - Root directory for search (default: current)

Detailed file operations guide →

💻 System Operations

Tools for executing commands and interacting with the system.

agent_run_shell_command(command, cwd=None, timeout=60)

Execute a shell command with streaming output, timeout handling, and safety confirmations.

Parameters:

command - The shell command to execute
cwd - Working directory (optional)
timeout - Inactivity timeout in seconds (default: 60)

Returns: success, stdout, stderr, exit_code, execution_time

🤖 Agent Operations

Tools for agent reasoning and sub-agent invocation.

agent_share_your_reasoning(reasoning, next_steps=None)

Share the agent's current thought process and planned actions. Essential for transparency and debugging.

Parameters:

reasoning - Current thought process and analysis
next_steps - Planned upcoming actions (optional)

list_agents()

List all available agents that can be invoked.

invoke_agent(agent_name, prompt, session_id=None)

Invoke a specific sub-agent with a given prompt. Useful for delegating specialized tasks.

Parameters:

agent_name - Name of the agent to invoke
prompt - The prompt to send to the agent
session_id - Optional session ID for conversation memory

🌐 Browser Automation Tools

Available to the qa-kitten agent for browser automation.

🐱 QA Kitten Only

These tools are only available when using the qa-kitten agent. Switch with /agent qa-kitten.

Browser Control

browser_initialize - Start a browser session
browser_close - Close the browser
browser_status - Check browser status

Navigation

browser_navigate - Navigate to a URL
browser_go_back - Go back in history
browser_go_forward - Go forward in history
browser_reload - Reload the page

Interactions

browser_click - Click an element
browser_set_text - Type text into an input
browser_get_text - Get text from an element
browser_select - Select from dropdown
browser_check - Check/uncheck checkbox

Element Location

browser_find_by_role - Find by ARIA role
browser_find_by_text - Find by text content
browser_find_by_label - Find by label

Visual

browser_screenshot_analyze - Take and analyze screenshots

Full browser tools reference →

Tools by Agent

Agent	Tools
`code-puppy`	All file ops, shell, grep, agent tools
`planning-agent`	File ops, reasoning tools
`code-reviewer`	File ops, grep, reasoning
`qa-kitten`	All above + 30+ browser tools
`agent-creator`	File ops (for creating agent configs)