HiveMind Node

Local AI gateway with multi-provider authentication.

Installation

npm install -g @hivemind/node

Or link locally during development:

cd packages/hivemind-node
npm link

Quick Start

# Interactive setup (recommended)
hivemind setup

This walks you through:

1. Select providers — Anthropic, OpenAI, DeepSeek, Google, Ollama
2. Authenticate — API key or browser session
3. See models — Shows available models per provider
4. Set default — Pick your primary provider
5. Connect — Optionally link to HiveMind server

Commands

# Setup & Status
hivemind setup              # Interactive setup wizard
hivemind status             # Show configured providers

# Models
hivemind models             # List available models
hivemind models -p anthropic  # Filter by provider

# Chat
hivemind chat "Hello"       # Quick chat with default provider
hivemind chat -p openai "Hi"  # Use specific provider
hivemind chat -m gpt-4o "Hi"  # Use specific model

# Provider Management
hivemind auth add <provider>    # Add a provider
hivemind auth list              # List providers
hivemind auth remove <provider> # Remove a provider
hivemind default <provider>     # Set default provider

# HiveMind Connection
hivemind connect <url>      # Connect to HiveMind server

Authentication Methods

API Key

Direct API access. Pay per token.

hivemind auth add anthropic
# Then enter your API key when prompted

Or with the key directly:

hivemind auth add anthropic -k sk-ant-xxx

Browser Session (Subscription)

Use your existing Claude Max ($20/mo) or ChatGPT Plus subscription.

During hivemind setup, choose "Browser Session" and paste your session cookie from DevTools:

1. Go to claude.ai (logged in)
2. DevTools → Application → Cookies
3. Copy sessionKey value

Environment Variables

The CLI auto-detects these:

export ANTHROPIC_API_KEY=sk-ant-xxx
export OPENAI_API_KEY=sk-xxx
export DEEPSEEK_API_KEY=sk-xxx

Supported Providers

| Provider | Models | Auth Methods | |----------|--------|--------------| | Anthropic | claude-opus-4, claude-sonnet-4, claude-3.5-* | API Key, Session | | OpenAI | gpt-4o, gpt-4-turbo, o1-* | API Key, Session | | DeepSeek | deepseek-chat, deepseek-coder | API Key | | Google | gemini-2.0-flash, gemini-1.5-pro | API Key | | Ollama | llama3, mistral, codellama | Local (no auth) |

How It Works

1. Encrypted Storage — Credentials in ~/.hivemind/auth.enc (AES-256-GCM)
2. Auto-Detection — Checks env vars and existing sessions
3. Health Checks — Validates credentials before use
4. HiveMind Integration — Reports capabilities for routing

HiveMind Connection

When connected to a HiveMind server, your node becomes an execution endpoint:

hivemind connect https://your-hivemind.vercel.app

The server can route LLM requests to your node based on:

• Which providers you have configured
• Which models you have access to
• Your node's availability

Credentials stay local — only capability info is shared with the server.

Development

cd packages/hivemind-node
npm install
npm run build
npm link

# Test
hivemind status

License

MIT