GLM-Vision MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ZHIPU_API_KEY | Yes | Your Zhipu (BigModel) API key for the GLM-4.6V-Flash vision model. |
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 | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| visionA | 识别一张图片，返回结构化文字描述与 OCR 文字。 适用于截图、文档、图表、照片等。当主文本模型无法直接读取图片、 或需要把图片内容转成文字再交给纯文本模型推理时，调用本工具。 参数: image_path: 图片的本地绝对路径（支持 png/jpg/jpeg/webp 等常见格式）。 prompt: 可选，自定义识别指令；不填则使用默认的"描述+OCR"指令。 |
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 1 tool
With only one tool, there is no possibility of overlap or misselection between tools. The single 'vision' tool has an unambiguous purpose (image recognition returning structured description and OCR text).
A single tool named 'vision' is trivially consistent since there are no siblings to clash with. It is a noun rather than a verb_noun pattern, but it is clear and readable.
One tool is thin in absolute terms, but the server has a tightly scoped single purpose (image-to-text), so a single capability fully matches its scope. It is appropriately minimal rather than under-built for the stated domain.
The tool covers the core need of converting an image to text with an optional custom prompt, which satisfies the surface for a vision utility. Minor gaps exist (no batch/multi-image handling or URL input), but agents can work around these by calling per-image.