MCP Image Validator

Python MCP (Model Context Protocol) Qwen3-VL Ollama Cloud.

?

MCP , Claude Code MCP vision Qwen3-VL (235B ) Ollama Cloud.

** :**

• ** ** - Qwen3-VL (235B )
• ** ** - Ollama Cloud API ( GPU)
• **MCP ** - Claude Code
• ** ** - JPEG, PNG, GIF, WebP, BMP
• **** - FastMCP
• **** -

---

• Python 3.10
• API Ollama Cloud ( )

#  
git clone <url->
cd mcp-image-validator

#  
pip install -r requirements.txt

#  
cp .env.example .env

#  .env    API 
#  : https://ollama.com/settings/keys

.env:

OLLAMA_API_KEY=_api__
VISION_MODEL=qwen3-vl:235b-cloud
OLLAMA_BASE_URL=https://ollama.com/v1
VISION_TEMPERATURE=0.2
VISION_MAX_TOKENS=1000

.env :

python test_full.py

:

• Ollama Cloud
  

•   (test.jpg, test.png, sample.jpg, sample.png)
  

•  Qwen3-VL
  

,       .

Claude Code

MCP Claude Code (.claude/mcp_settings.json claude_desktop_config.json):

{
  "mcpServers": {
    "image-validator": {
      "command": "python",
      "args": ["c:\\GIT\\mcp-image-validator\\server.py"],
      "env": {
        "OLLAMA_API_KEY": "_api__"
      }
    }
  }
}

:

• API env

• Claude Code

  Claude Code:

    C:\Users\user\Pictures\photo.jpg

---

describe_image

    Qwen3-VL.

:

• image_path (, ) -
• prompt (, ) -

** :**

    C:\Photos\landscape.jpg

  C:\Docs\diagram.png    

    /home/user/pictures/sunset.jpg?

** :** JPEG, PNG, GIF, WebP, BMP

---

OLLAMA_API_KEY	-	. API Ollama Cloud
VISION_MODEL	qwen3-vl:235b-cloud	Vision
OLLAMA_BASE_URL	https://ollama.com/v1	Ollama Cloud API
VISION_TEMPERATURE	0.2	(0.0-1.0, = )
VISION_MAX_TOKENS	1000

: Ollama Cloud -cloud (, qwen3-vl:235b-cloud).

---

FastMCP ( Anthropic) OpenAI SDK, Ollama Cloud.

:

• MCP Server (server.py) - FastMCP stdio
• Ollama Client (ollama_vision_client.py) - API
• Qwen3-VL Model - Vision 235B Ollama Cloud

** :**

Claude Code  MCP Protocol  server.py  ollama_vision_client.py
                                              
                                         OpenAI SDK
                                              
                                    Ollama Cloud API
                                              
                                       Qwen3-VL Model

** :**

• comfyui-mcp-server - MCP
• doc-chatbot - Ollama Cloud

---

mcp-image-validator/
 server.py                   # MCP  (stdio transport)
 ollama_vision_client.py     #  Ollama Cloud API
 test_full.py                #   
 requirements.txt            # Python 
 .env.example                #   
 .gitignore                  #   git
 README.md                   #  

---

python server.py

 MCP    stdin.

#   
python test_full.py

#   
python test_full.py
#       

stderr.    Claude Code     MCP .

  `server.py`:

logging.basicConfig(level=logging.DEBUG)

---

"OLLAMA_API_KEY not set"

:

• , .env
• Claude Code env MCP

"Image file not found"

:

• : dir "C:\path\to\image.jpg" (Windows) ls -la /path/to/image.jpg (Linux/Mac)

"Unsupported image format"

:

• : JPEG, PNG, GIF, WebP, BMP
• ,

(10-30+ )

** !**

• Vision 235B

Claude Code

:

• MCP
• python ( python3 )
• Claude Code
• MCP

"Connection refused" "Network error"

:

• ollama.com
• API

---

    C:\Users\user\Pictures\photo.jpg

  C:\Docs\screenshot.png      

  /home/user/vacation.jpg:
1.   
2.   ?
3.    ?

Claude Code describe_image .

---

• Python 3.10
• ( Ollama Cloud API)
• API Ollama Cloud

• mcp 1.0.0 - FastMCP
• openai 1.0.0 - OpenAI SDK ( Ollama Cloud)
• python-dotenv 1.0.0 -
• Pillow 10.0.0 -

:

pip install -r requirements.txt

• ** :** 10-30 ( )
• : ( 235B )
• : Production-ready

---

:

@mcp.tool()
def analyze_colors(image_path: str) -> dict:
    """   """
    #  
    pass

VISION_MODEL .env:

#  vision    -cloud
VISION_MODEL=llava:7b-cloud
VISION_MODEL=bakllava:7b-cloud

---

! :

• (OCR,  )
  

• vision
• Unit

Issues Pull Requests GitHub.

---

• README.md - ( )
• test_full.py -

---

MIT License - . LICENSE .

---

• Ollama Cloud - AI
• Ollama API Keys - API
• MCP Protocol - Model Context Protocol
• Claude Code - AI
• FastMCP - MCP

---

:

1. (#-)
2. Issue GitHub :

• Python (python --version)

---