Skip to content

Configuration Guide

Environment Variables

All configuration is done through environment variables. Create a .env file or set system environment variables directly.

Anthropic API Configuration

Required Configuration:

ANTHROPIC_API_KEY=sk-ant-xxxxxxxxxxxxx

Optional Configuration (for custom API endpoints):

ANTHROPIC_BASE_URL=https://api.anthropic.com
ANTHROPIC_API_VERSION=2023-06-01

MiniMax Configuration

If using MiniMax as a model provider, configure as follows:

ANTHROPIC_BASE_URL=https://api.minimaxi.chat/v1
ANTHROPIC_API_KEY=your_minimax_api_key

Enable Serper.dev as a web search engine:

SERPER_API_KEY=your_serper_api_key

Note: If not configured, Anthropic's native search functionality will be used.

Disable Interleaved Thinking

To disable the model's interleaved thinking feature:

ANTHROPIC_DISABLE_INTERLEAVED_THINKING=true

Other Configuration

# Runtime Environment
NODE_ENV=development|production

# Log Level
LOG_LEVEL=debug|info|warn|error

# Debug Mode
DEBUG=*

Configuration Priority

Environment variable loading priority (from high to low):

  1. System Environment Variables - Variables set directly in the system
  2. .env File - .env file in project root
  3. .env.local - Local environment file (not committed to Git)
  4. Default Values - Default values in code

Configuration File Locations

.env File

Create a .env file in the project root:

open-claude-code/
├── .env                 # Main configuration file
├── .env.local          # Local overrides (not committed)
├── .env.example        # Example configuration (committed)
└── ...

.gitignore Configuration

Ensure the following files are in .gitignore to avoid committing sensitive configurations:

.env
.env.local
.env.*.local

Common Configuration Scenarios

Scenario 1: Local Development

ANTHROPIC_API_KEY=your_api_key
NODE_ENV=development
LOG_LEVEL=debug

Scenario 2: Production Environment

ANTHROPIC_API_KEY=your_prod_api_key
NODE_ENV=production
LOG_LEVEL=info
SERPER_API_KEY=your_serper_key

Scenario 3: MiniMax + Serper

ANTHROPIC_BASE_URL=https://api.minimaxi.chat/v1
ANTHROPIC_API_KEY=minimax_key
SERPER_API_KEY=serper_key
NODE_ENV=production

Scenario 4: Offline Mode (Local Tools Only)

ANTHROPIC_API_KEY=placeholder_key
# Use only local tools like file operations, Git, etc.

Verify Configuration

1. Check Environment Variables

# Display all configured Anthropic-related variables
env | grep ANTHROPIC

2. Test API Connection

bun run cli advisor "test connection"

3. View Configuration Logs

Enable debug mode to view configuration loading process:

DEBUG=* bun run cli --version

Troubleshooting

Issue: Invalid API Key

Solution: 1. Verify the key is correctly copied 2. Check if the key has expired 3. Ensure the key has proper permissions 4. Review API documentation to confirm key format

Issue: Web Search Not Working

Solution: 1. Verify SERPER_API_KEY is configured 2. Check if API quota is exhausted 3. If using custom URL, check network connection

Issue: Slow Model Response

Solution: 1. Check network connection 2. If using custom endpoint, verify endpoint availability 3. Try disabling interleaved thinking: ANTHROPIC_DISABLE_INTERLEAVED_THINKING=true

Best Practices

  1. Don't Commit Sensitive Configurations
  2. Use .env.local to store local overrides

  3. Ensure .env and .env.local are in .gitignore

  4. Use Configuration Examples

  5. Commit .env.example as a configuration template

  6. New developers can start from the example file

  7. Environment Isolation

  8. Use different API keys for different environments (development, testing, production)

  9. Separate configurations using .env.development, .env.test, etc.

  10. Document Configuration Changes

  11. Record new configuration options in CHANGELOG
  12. Update configuration instructions in documentation

Next Steps