VisionToolMCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| GEMINI_API_KEY | No | Gemini API key. | |
| OPENAI_API_KEY | No | OpenAI API key. | |
| VISIONTOOL_MODEL | No | Model to use. Defaults: claude-opus-4-8 (anthropic), gpt-4o-mini (openai), gemini-3.5-flash (gemini). | |
| ANTHROPIC_API_KEY | No | Anthropic API key. | |
| VISIONTOOL_API_KEY | No | Unified API key for any provider. Alternative to provider-specific keys. | |
| VISIONTOOL_RETRIES | No | Number of retries on temporary API failures. | 2 |
| VISIONTOOL_BASE_URL | No | Base URL for the API. | |
| VISIONTOOL_API_FORMAT | No | API format: anthropic, openai, or gemini. | anthropic |
| VISIONTOOL_TIMEOUT_MS | No | Request timeout in milliseconds. | 60000 |
| VISIONTOOL_RETRY_BASE_MS | No | Base delay in milliseconds for exponential backoff. | 250 |
| VISIONTOOL_MAX_IMAGE_BYTES | No | Maximum image size in bytes for local/base64 images. | 5242880 |
| VISIONTOOL_ALLOWED_CALLER_PREFIXES | No | Comma-separated list of allowed caller model prefixes. | glm,deepseek |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| describe_imageC | 【仅限 GLM/DeepSeek 系列模型调用】Describe an image for a text-only agent. Accepts a local path, base64 image data, or URL. |
| ocr_imageB | 【仅限 GLM/DeepSeek 系列模型调用】Extract visible text from an image with optional language and layout hints. |
| answer_about_imageA | 【仅限 GLM/DeepSeek 系列模型调用】Answer a specific question using visual evidence from one image. |
| compare_imagesA | 【仅限 GLM/DeepSeek 系列模型调用】Compare two images and summarize relevant differences for a text-only agent. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 4 tools
Each tool targets a distinct capability: answering specific questions, comparing two images, describing content, and extracting text. There is no overlap in their purposes.
All tools use snake_case with verb-first pattern (answer, compare, describe, ocr). However, 'answer_about_image' uses a preposition while others directly combine verb and noun, a minor inconsistency.
Four tools is appropriate for a focused vision server providing core image understanding capabilities. Not too few or too many.
Covers key image interpretation needs: description, comparison, OCR, and question answering. Missing potential features like object detection or image generation, but the set is reasonably complete for its stated purpose of supporting text-only agents.