Configuration
Docxa uses environment variables for configuration. You can manage these via a .env file in your workspace root or by passing them directly to your shell.
Environment Variables
| Variable | Description | Default |
|---|---|---|
DOCXA_PROVIDER |
LLM Provider (openai, anthropic, google-gemini, azure-openai, ollama) |
openai |
DOCXA_MODEL |
Specific model to use (e.g., gpt-4o, llama3.1) |
Provider-specific |
DOCXA_API_KEY |
Overwrites any provider-specific API keys | None |
OPENAI_API_KEY |
Required if provider is openai |
None |
ANTHROPIC_API_KEY |
Required if provider is anthropic |
None |
GEMINI_API_KEY |
Required if provider is google-gemini |
None |
AZURE_OPENAI_API_KEY |
Required if provider is azure-openai |
None |
AZURE_OPENAI_ENDPOINT |
Azure OpenAI resource endpoint | None |
AZURE_OPENAI_DEPLOYMENT |
Azure deployment/model name | None |
AZURE_OPENAI_API_MODE |
Azure API mode (v1 or legacy) |
v1 |
AZURE_OPENAI_API_VERSION |
Required only for legacy mode |
2024-10-21 |
DOCXA_OLLAMA_URL |
Custom endpoint for Ollama | http://localhost:11434/v1 |
Local .env File
Docxa automatically loads environment variables from:
.env.local.env
If DOCXA_PROVIDER is omitted, Docxa will auto-detect a provider from the available credentials, preferring Azure OpenAI, then OpenAI, Anthropic, Gemini, and finally Ollama.
Example .env:
DOCXA_PROVIDER=openai
OPENAI_API_KEY=sk-proj-xxxxxx
Azure example:
DOCXA_PROVIDER=azure-openai
AZURE_OPENAI_API_KEY=your-azure-key
AZURE_OPENAI_ENDPOINT=https://your-resource-name.cognitiveservices.azure.com/
AZURE_OPENAI_DEPLOYMENT=gpt-5-mini
AZURE_OPENAI_API_MODE=v1
DOCXA_MODEL=gpt-5-mini
Custom Env File
You can specify a custom environment file using the --env-file flag:
docxa generate prd --env-file .env.production