Quick Start Guide
Quick Start Guide
Get your AI coding agent connected to an LLM provider in minutes.
The 4-Step Process
1. Choose a Provider
Pick an LLM provider based on your needs. See the Provider Overview for a comparison, or jump straight to a setup guide:
- Google Gemini — Fast, generous free tier
- Anthropic Claude — Best reasoning and code quality
- OpenAI GPT — Broad ecosystem support
- DeepSeek — Budget-friendly, strong coding
- Groq — Ultra-fast inference
2. Get an API Key
Sign up with your chosen provider and generate an API key:
| Provider | Sign Up |
|---|---|
| Gemini | aistudio.google.com |
| Claude | console.anthropic.com |
| OpenAI | platform.openai.com |
| DeepSeek | platform.deepseek.com |
| Groq | console.groq.com |
3. Set Your Environment Variable
Export the key so your agent can access it:
export GEMINI_API_KEY="your-key-here"Persist it across sessions:
echo 'export GEMINI_API_KEY="your-key"' >> ~/.bashrcsource ~/.bashrc4. Configure Your Agent
Connect your provider to an AI coding agent:
- OpenCode — CLI agent, JSON config
- Claude Desktop — GUI app
- Cursor / Windsurf — IDE integrations
- VS Code Copilot — Editor extension
Verify Your Setup
-
Check the key is set:
Terminal window echo $GEMINI_API_KEY -
Confirm your config file is valid JSON:
Terminal window cat opencode.json | python3 -m json.tool -
Run your agent and send a test prompt. If you get a response, you’re set.
Next Steps
- Choosing a Provider — Detailed comparison
- Troubleshooting — Fix common problems