z_ai_vision_mcp_server_clone
by 1orZero
README.md
# z_ai_vision_mcp_server_clone
OpenAI-compatible MCP server for running image analysis tools against your own vision model endpoint.
## Tools
- `ui_to_artifact`
- `extract_text_from_screenshot`
- `diagnose_error_screenshot`
- `understand_technical_diagram`
- `analyze_data_visualization`
- `ui_diff_check`
- `analyze_image`
## Configuration
Set either `VISION_ENDPOINT` or `VISION_BASE_URL`.
| Variable | Required | Description |
| --- | --- | --- |
| `VISION_ENDPOINT` | Yes, unless `VISION_BASE_URL` is set | Full chat completions endpoint. |
| `VISION_BASE_URL` | Yes, unless `VISION_ENDPOINT` is set | Base URL; `/chat/completions` is appended. |
| `VISION_MODEL` | Yes | Vision model name sent in the request body. |
| `VISION_API_KEY` | No | Bearer token. Omit for local endpoints that do not require auth. |
| `VISION_PROVIDER` | No | Label for your provider. Defaults to `custom`. |
| `VISION_MAX_IMAGE_MB` | No | Local image size limit. Defaults to `5`. |
| `VISION_TIMEOUT_MS` | No | Request timeout. Defaults to `300000`. |
| `VISION_TEMPERATURE` | No | Optional model temperature. |
| `VISION_TOP_P` | No | Optional model top_p. |
| `VISION_MAX_TOKENS` | No | Optional max_tokens. |
You can also place these values in a local `.env` file in the working directory where the server starts. Real environment variables override `.env` values.
## Run
```bash
npm install
npm run build
VISION_ENDPOINT=http://localhost:11434/v1/chat/completions VISION_MODEL=llava npm start
```
Or with `.env`:
```bash
npm start
```
## MCP Client Example
```json
{
"mcpServers": {
"z-ai-vision-clone": {
"type": "stdio",
"command": "npx",
"args": ["-y", "z_ai_vision_mcp_server_clone"],
"env": {
"VISION_ENDPOINT": "https://your-provider.com/v1/chat/completions",
"VISION_MODEL": "your-vision-model",
"VISION_API_KEY": "your-api-key"
}
}
}
}
```
TDQS
B3/5.0
Scored across 7 tools
Disambiguation4/5
Each tool has a distinct purpose (e.g., data visualization vs. error screenshot vs. text extraction), but 'analyze_image' as a catch-all creates slight ambiguity since it overlaps with the specialized tools.
Naming Consistency4/5
Most names follow a verb_noun pattern (analyze_*, extract_text_*, understand_*), but 'ui_diff_check' and 'ui_to_artifact' deviate slightly from the typical structure.
Tool Count5/5
With 7 tools focused on image and UI analysis, the count is well-scoped—no unnecessary duplication and full coverage of the domain.
Completeness5/5
The tool set covers major visual analysis tasks (charts, screenshots, text extraction, UI comparison, diagram understanding) with no obvious gaps for the intended use case.
Maintenance
ActivityInactive
ResponsivenessNo recent community issues