Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OSS_BUCKET | No | OSS/S3 bucket name (optional, required when GENAI_IMAGE_FORMAT=url) | |
| OSS_REGION | No | OSS/S3 region (optional, required when GENAI_IMAGE_FORMAT=url) | us-east-1 |
| SERVER_PORT | No | HTTP server port | 8080 |
| OSS_ENDPOINT | No | OSS/S3 endpoint (optional, required when GENAI_IMAGE_FORMAT=url). For AWS S3: leave empty or set to s3.amazonaws.com. For Aliyun OSS: set to oss-cn-hangzhou.aliyuncs.com or your region. For Tencent COS: set to cos.ap-guangzhou.myqcloud.com. For MinIO: set to your MinIO endpoint | |
| GENAI_API_KEY | Yes | Your API key (Google Gemini API key, dmxapi key, or DashScope API key depending on provider) | |
| GENAI_BASE_URL | No | Shared GenAI endpoint for both providers | https://generativelanguage.googleapis.com |
| GENAI_PROVIDER | No | GenAI provider: 'gemini' for Google Gemini / compatible backend, 'wan' for Ali Bailian Tongyi Wanxiang image APIs | gemini |
| OSS_ACCESS_KEY | No | OSS/S3 access key (optional, required when GENAI_IMAGE_FORMAT=url) | |
| OSS_SECRET_KEY | No | OSS/S3 secret key (optional, required when GENAI_IMAGE_FORMAT=url) | |
| SERVER_ADDRESS | No | HTTP server address | 0.0.0.0 |
| GENAI_IMAGE_FORMAT | No | Image output format: 'base64' to return image as data URI (base64 encoded), 'url' to upload image to OSS and return plain URL | base64 |
| GENAI_GEN_MODEL_NAME | No | Model name for image generation. When GENAI_PROVIDER=gemini: Gemini model names (e.g. gemini-3-pro-image-preview). When GENAI_PROVIDER=wan: Wanxiang model names (e.g. wan2.5-t2i-preview) | gemini-3-pro-image-preview |
| GENAI_EDIT_MODEL_NAME | No | Model name for image editing. When GENAI_PROVIDER=gemini: Gemini model names (e.g. gemini-3-pro-image-preview). When GENAI_PROVIDER=wan: Wanxiang model names (e.g. wan2.5-i2i-preview) | gemini-3-pro-image-preview |
| GENAI_TIMEOUT_SECONDS | No | Request timeout in seconds for each GenAI call (generate / edit) | 120 |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
No tools | |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |