sidecar

Sidecar: Multi-Model Subagent Tool

Spawn parallel conversations with different LLMs (Gemini, GPT, ChatGPT, Codex, o3, etc.) and fold results back into your context.

Installation

npm install -g claude-sidecar

Verify installation:

sidecar --version

Requirements

• Node.js 18+
• API credentials (see Setup below)
• Electron (installed automatically as optional dependency)

MCP Server (Auto-Registered)

On install, an MCP server is auto-registered for Claude Cowork and Claude Desktop. If you're in an MCP-enabled environment, you can use sidecar_start, sidecar_status, sidecar_read, and other MCP tools directly instead of CLI commands. Call sidecar_guide for detailed usage instructions.

---

Setup: Configuring API Access

Sidecar uses the OpenCode SDK to communicate with LLM providers. You need to configure API credentials for your chosen provider(s).

Option A: OpenRouter (Recommended for Multi-Model Access)

OpenRouter provides unified access to many models (Gemini, GPT-4, Claude, o3, etc.) with a single API key.

Step 1: Get an OpenRouter API key

• Sign up at https://openrouter.ai
• Go to Keys → Create Key
• Copy your key (starts with sk-or-v1-...)

Step 2: Configure credentials

Create the auth file:

mkdir -p ~/.local/share/opencode
cat > ~/.local/share/opencode/auth.json << 'EOF'
{
  "openrouter": {
    "apiKey": "sk-or-v1-YOUR_KEY_HERE"
  }
}
EOF

Step 3: Verify setup

sidecar start --model gemini --prompt "Say hello" --no-ui

Model names with OpenRouter: When using OpenRouter, prefix the model with openrouter/:

sidecar start --model gemini --prompt "..."
sidecar start --model gpt --prompt "..."
sidecar start --model claude --prompt "..."

Option B: Direct API Keys (Provider-Specific)

Use this if you have API keys directly from Google, OpenAI, or Anthropic.

For Google Gemini:

export GOOGLE_GENERATIVE_AI_API_KEY=your-google-api-key

For OpenAI:

export OPENAI_API_KEY=your-openai-api-key