# @lov3kaizen/agentsea-cli **Command-line interface for AgentSea ADK** - Build and orchestrate AI agents from your terminal. [![npm version](https://img.shields.io/npm/v/@lov3kaizen/agentsea-cli.svg)](https://www.npmjs.com/package/@lov3kaizen/agentsea-cli) [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT) ## Features - 🚀 **Quick Setup** - Initialize with interactive prompts - 💬 **Interactive Chat** - Chat with agents in your terminal - 💻 **Agentic Coding** - AI coding assistant with 13 built-in tools (file ops, git, shell, search) - 🤖 **Agent Management** - Create, list, and manage agents - 🔌 **Provider Support** - Cloud and local providers - 📦 **Model Management** - Pull and manage Ollama models - ⚙️ **Configuration** - Persistent configuration management - 🎨 **Beautiful Output** - Colored, formatted terminal output with streaming ## Installation ```bash # Global installation npm install -g @lov3kaizen/agentsea-cli # Or use with npx npx @lov3kaizen/agentsea-cli init ``` ## Quick Start ### 1. Initialize ```bash agentsea init ``` This will guide you through: - Choosing a provider (cloud or local) - Configuring API keys or endpoints - Creating a default agent ### 2. Start Chatting ```bash agentsea chat ``` Interactive chat session with your default agent. ### 3. Start Coding ```bash agentsea code ``` Interactive agentic coding session with file operations, git, search, and shell tools. ### 4. Run One-Off Commands ```bash agentsea agent run default "What is the capital of France?" ``` ## Commands ### `agentsea code` Start an interactive agentic coding session with 13 built-in tools. ```bash agentsea code # Use default agent agentsea code --agent my-agent # Use specific agent agentsea code --provider anthropic # Override provider agentsea code --model claude-sonnet-4-6 # Override model agentsea code --verbose # Show tool calls, tokens, latency agentsea code --maxIterations 50 # Limit tool iterations (default: 25) ``` **Built-in Coding Tools (13 tools):** | Tool | Description | | --------------- | --------------------------------------- | | `file_read` | Read file contents | | `file_write` | Write content to files | | `file_list` | List directory contents | | `code_edit` | Precise search-and-replace editing | | `glob` | Find files matching glob patterns | | `grep` | Search file contents with regex | | `shell_execute` | Run shell commands (with safety checks) | | `git_status` | Show working tree status | | `git_diff` | Show changes between commits | | `git_add` | Stage files for commit | | `git_commit` | Create a commit | | `git_log` | Show commit history | | `git_branch` | List or create branches | The coding agent follows best practices: reads files before editing, uses precise edits over rewrites, and verifies changes with git diff. ### `agentsea init` Initialize AgentSea CLI configuration with interactive prompts. ```bash agentsea init ``` ### `agentsea chat` Start an interactive chat session. ```bash agentsea chat # Use default agent agentsea chat --agent my-agent # Use specific agent agentsea chat --model llama3 # Override model ``` ### `agentsea agent` Manage agents. ```bash # Create a new agent agentsea agent create # List all agents agentsea agent list # Get agent details agentsea agent get # Run an agent with a message agentsea agent run "Your message" # Set default agent agentsea agent default # Delete an agent agentsea agent delete ``` ### `agentsea provider` Manage providers. ```bash # List all providers agentsea provider list # Get provider details agentsea provider get # Add a new provider agentsea provider add # Set default provider agentsea provider default # Delete a provider agentsea provider delete ``` ### `agentsea model` Manage models (Ollama only). ```bash # List available models agentsea model list # Pull a model from Ollama agentsea model pull llama2 # Show popular models agentsea model popular ``` ### `agentsea config` Show current configuration. ```bash agentsea config ``` ## Examples ### Cloud Provider (Anthropic) ```bash # Initialize with Anthropic agentsea init > Cloud Provider > Anthropic > [Enter API Key] # Chat with Claude agentsea chat ``` ### Local Provider (Ollama) ```bash # Initialize with Ollama agentsea init > Local Provider > Ollama > http://localhost:11434 # Pull a model agentsea model pull llama2 # Chat with local model agentsea chat ``` ### Multiple Agents ```bash # Create a coding assistant agentsea agent create > Name: coder > Model: codellama > System Prompt: You are a coding assistant... # Create a writer assistant agentsea agent create > Name: writer > Model: llama2 > System Prompt: You are a creative writer... # Use specific agent agentsea chat --agent coder ``` ## Configuration Configuration is stored in: - **Linux**: `~/.config/agentsea-cli/config.json` - **macOS**: `~/Library/Preferences/agentsea-cli/config.json` - **Windows**: `%APPDATA%\agentsea-cli\config.json` ### Configuration Structure ```json { "defaultProvider": "anthropic", "defaultAgent": "default", "providers": { "anthropic": { "name": "anthropic", "type": "anthropic", "apiKey": "sk-ant-...", "timeout": 60000 }, "ollama": { "name": "ollama", "type": "ollama", "baseUrl": "http://localhost:11434", "timeout": 60000 } }, "agents": { "default": { "name": "default", "description": "Default agent", "model": "claude-sonnet-4-6", "provider": "anthropic", "systemPrompt": "You are a helpful assistant.", "temperature": 0.7, "maxTokens": 2048 } } } ``` ### Environment Variables API keys can also be set via environment variables: ```bash export ANTHROPIC_API_KEY=your_key_here export OPENAI_API_KEY=your_key_here export GEMINI_API_KEY=your_key_here ``` ## Supported Providers ### Cloud Providers - **Anthropic** - Claude models - **OpenAI** - GPT models - **Google** - Gemini models ### Local Providers - **Ollama** - Local LLM runtime - **LM Studio** - GUI for local models - **LocalAI** - OpenAI-compatible local API ## Ollama Integration The CLI has first-class support for Ollama: ```bash # Make sure Ollama is running ollama serve # Pull popular models agentsea model pull llama2 agentsea model pull mistral agentsea model pull codellama # List available models agentsea model list # Show popular models agentsea model popular # Chat with local model agentsea chat ``` ## Tips & Tricks ### 1. Quick Chat Create an alias for quick access: ```bash alias bc="agentsea chat" ``` ### 2. Multiple Providers Set up multiple providers for different use cases: ```bash agentsea provider add > Name: anthropic-prod > Type: Anthropic agentsea provider add > Name: ollama-dev > Type: Ollama ``` ### 3. Specialized Agents Create agents for specific tasks: ```bash # Coding agent agentsea agent create > Name: code > Model: codellama > System Prompt: You are an expert programmer... # Writing agent agentsea agent create > Name: write > Model: llama2 > System Prompt: You are a creative writer... # Use them agentsea chat --agent code agentsea chat --agent write ``` ### 4. Verbose Mode Get detailed output: ```bash agentsea agent run default "Hello" --verbose ``` ## Troubleshooting ### "No providers configured" Run `agentsea init` to set up your first provider. ### "Provider not found" List providers: `agentsea provider list` Add provider: `agentsea provider add` ### "Agent not found" List agents: `agentsea agent list` Create agent: `agentsea agent create` ### Ollama Connection Error Make sure Ollama is running: ```bash ollama serve ``` Check the base URL in your provider configuration: ```bash agentsea provider get ollama ``` ### Model Not Found (Ollama) Pull the model first: ```bash agentsea model pull llama2 ``` ## Development ```bash # Install dependencies pnpm install # Build the CLI pnpm build # Link for local testing pnpm link --global # Test commands agentsea--help ``` ## License MIT License - see [LICENSE](../../LICENSE) for details ## See Also - [@lov3kaizen/agentsea-core](../core) - Core AgentSea ADK - [@lov3kaizen/agentsea-nestjs](../nestjs) - NestJS integration - [AgentSea Documentation](../../docs) --- Built with ❤️ by [lovekaizen](https://lovekaizen.com)