AgentSea - Unite and Orchestrate AI Agents

Built for Developer Productivity

✅ Interactive setup wizard

✅ Real-time chat interface

✅ Agent management (CRUD operations)

✅ Model management (pull, list, info)

✅ Configuration persistence

✅ Voice support

Installation

Install the AgentSea CLI globally with npm:

bash

npm install -g @lov3kaizen/agentsea-cli
pnpm add -g @lov3kaizen/agentsea-cli
yarn global add @lov3kaizen/agentsea-cli
sea --version

Quick Start

Initialize AgentSea with the interactive setup wizard:

bash

sea init
# The wizard will guide you through:
# 1. Choose your provider (Anthropic, OpenAI, Ollama, etc.)
# 2. Enter API key (if cloud provider)
# 3. Select default model
# 4. Configure voice providers (optional)
# 5. Save configuration

Configuration

Configuration is stored in:

macOS: ~/Library/Preferences/agentsea-cli
Linux: ~/.config/agentsea-cli
Windows: %APPDATA%/agentsea-cli

Commands

sea init

Interactive initialization wizard to set up AgentSea configuration.

bash

sea init
Choose your provider: (Use arrow keys)
Anthropic (Claude)
OpenAI (GPT)
Google (Gemini)
Ollama (Local)
LM Studio (Local)
LocalAI (Local)
Enter your API key: ****
Choose default model: claude-sonnet-4-20250514
Configure voice? Yes
STT provider: openai-whisper
TTS provider: openai-tts
Configuration saved!

sea chat

Start an interactive chat session with your agent.

bash

sea chat
sea chat --agent my-assistant
sea chat --voice
sea chat --model llama3.2
AgentSea Chat (Type 'exit' to quit)
Hello! What can you help me with?
I'm a helpful AI assistant. I can help with:
- Answer questions
- Write code
- Analyze data
- Creative writing
- And much more!
Calculate 42 * 58
42 × 58 = 2,436

sea agent

Manage your agents - create, list, get, delete, run.

Create an Agent

bash

sea agent create
sea agent create my-assistant   --provider anthropic   --model claude-sonnet-4-20250514   --system-prompt "You are a helpful coding assistant"   --temperature 0.7   --max-tokens 2048
sea agent create voice-assistant   --provider ollama   --model llama3.2   --voice-stt openai-whisper   --voice-tts openai-tts

List Agents

bash

sea agent list
Available Agents:
================
my-assistant (default)
  Provider: anthropic
  Model: claude-sonnet-4-20250514
voice-assistant
  Provider: ollama
  Model: llama3.2
  Voice: OpenAI Whisper (STT) + OpenAI TTS (TTS)
code-reviewer
  Provider: openai
  Model: gpt-4-turbo-preview

Run an Agent

bash

sea agent run my-assistant "What is 42 * 58?"
sea agent run voice-assistant   --voice-input ./question.mp3   --voice-output ./response.mp3
sea agent run my-assistant "Explain async/await" {'>'} explanation.txt

Other Agent Commands

bash

sea agent get my-assistant
sea agent delete my-assistant
sea agent default my-assistant

sea model

Manage local models (works with Ollama).

bash

sea model pull llama3.2
Pulling llama3.2...
████████████████████████████████ 100% (4.7 GB)
Model llama3.2 pulled successfully!
sea model list
Available Models:
================
llama3.2 (4.7 GB)
mistral (4.1 GB)
codellama (3.8 GB)
qwen2.5 (4.7 GB)
sea model info llama3.2
Model: llama3.2
Size: 4.7 GB
Parameters: 8B
Quantization: Q4_0
Family: LLaMA
License: Llama 3 Community License
sea model remove codellama

sea provider

Manage provider configurations.

bash

sea provider add anthropic --api-key sk-ant-...
sea provider list
Configured Providers:
====================
anthropic (Claude)
  API Key: sk-ant-****
  Default Model: claude-sonnet-4-20250514
ollama (Local)
  Base URL: http://localhost:11434
  Models: 4 installed
sea provider test anthropic
sea provider remove openai

Common Workflows

1. Quick Local Setup

bash

curl -fsSL https://ollama.com/install.sh | sh
sea init
# Select: Ollama (Local)
sea model pull llama3.2
sea chat
No API keys needed!
Complete privacy!
Zero costs!

2. Create Voice-Enabled Assistant

bash

sea agent create voice-bot   --provider anthropic   --model claude-sonnet-4-20250514   --voice-stt openai-whisper   --voice-tts openai-tts   --system-prompt "You are a voice assistant. Keep responses concise."
sea chat --agent voice-bot --voice
sea agent run voice-bot   --voice-input question.mp3   --voice-output answer.mp3

3. Multi-Agent Setup

bash

sea agent create researcher   --model claude-sonnet-4-20250514   --system-prompt "Research and gather information"
sea agent create writer   --model gpt-4-turbo-preview   --system-prompt "Write clear, engaging content"
sea agent create coder   --provider ollama   --model codellama   --system-prompt "Write clean, efficient code"
sea agent run researcher "Latest AI trends"
sea agent run writer "Blog post about AI agents"
sea agent run coder "Python function for data validation"

Configuration File

AgentSea stores configuration in JSON format. You can edit it directly if needed:

json

{
  "providers": {
    "anthropic": {
      "apiKey": "sk-ant-****",
      "defaultModel": "claude-sonnet-4-20250514"
    },
    "ollama": {
      "baseUrl": "http://localhost:11434"
    }
  },
  "agents": {
    "my-assistant": {
      "provider": "anthropic",
      "model": "claude-sonnet-4-20250514",
      "systemPrompt": "You are a helpful assistant.",
      "temperature": 0.7,
      "maxTokens": 2048
    },
    "local-agent": {
      "provider": "ollama",
      "model": "llama3.2",
      "temperature": 0.8
    }
  },
  "defaultAgent": "my-assistant",
  "voice": {
    "stt": {
      "provider": "openai-whisper",
      "apiKey": "sk-****"
    },
    "tts": {
      "provider": "openai-tts",
      "apiKey": "sk-****",
      "voice": "nova"
    }
  }
}

Tips & Best Practices

Use Specialized Agents

Create different agents for different tasks: code review, writing, data analysis, etc. Each can have optimized prompts and models.

Mix Cloud & Local

Use local models for development and simple tasks, cloud models for complex production work. Best of both worlds.

Keep Keys Secure

The CLI stores API keys securely in your system config. Never commit them to version control.

Voice Mode Tips

Keep voice assistant prompts concise. Add "Respond in 2 sentences max" for better voice UX. Users don't want to listen to long responses.

Complete Examples

Example 1: Local Development Setup

bash

curl -fsSL https://ollama.com/install.sh | sh
npm install -g @lov3kaizen/agentsea-cli
ollama pull llama3.2
sea init  # Choose Ollama
sea chat
# Start coding assistance with local AI
# No API costs, complete privacy!

Example 2: Production Setup

CODE

sea provider add anthropic --api-key sk-ant-***sea provider add openai --api-key sk-***sea agent create prod-assistant \ --provider anthropic \ --model claude-sonnet-4-20250514 \ --temperature 0.7 \ --max-tokens 2048sea agent default prod-assistantsea agent run prod-assistant "Analyze this error log..." < error.log

Example 3: Voice Assistant

bash

sea init
# Configure voice providers when prompted
sea agent create jarvis   --provider anthropic   --voice-stt openai-whisper   --voice-tts openai-tts
sea chat --agent jarvis --voice
sea agent run jarvis   --voice-input ~/Desktop/question.mp3   --voice-output ~/Desktop/answer.mp3

Next Steps

Voice Features - Learn more about voice capabilities
Local Models - Deep dive into local execution
Agent Concepts - Understand agent architecture
View Examples - Complete CLI examples

Pro Tip

Use sea chat --voice for a hands-free coding experience. Ask questions while looking at your code, get answers spoken back to you. Perfect for pair programming with AI!

CLI Tool

Built for Developer Productivity

Installation

Quick Start

Configuration

Commands

sea init

sea chat

sea agent

Create an Agent

List Agents

Run an Agent

Other Agent Commands

sea model

sea provider

Common Workflows

1. Quick Local Setup

2. Create Voice-Enabled Assistant

3. Multi-Agent Setup

Configuration File

Tips & Best Practices

Use Specialized Agents

Mix Cloud & Local

Keep Keys Secure

Voice Mode Tips

Complete Examples

Example 1: Local Development Setup

Example 2: Production Setup

Example 3: Voice Assistant

Next Steps

Pro Tip