mcp-vision-ollama mcp

v1.0.0 · MCP Tool · ai · registry.pascalai.org

Analyze images using local Ollama vision models (llava, bakllava, moondream, etc.).

ollamavisionlocalllavaimageanalysis

Input Parameters

Parameter	Type	Description
hostoptional	string	Default: `http://localhost:11434`.
modeloptional	string	Vision-capable model, e.g. llava, bakllava, moondream. Default: `llava`.
promptrequired	string	Default: `Describe this image.`.
image_pathoptional	string	Local image path.
image_base64optional	string	Base64 encoded image.
image_urloptional	string
systemoptional	string
temperatureoptional	number	Default: `0.7`.
max_tokensoptional	integer

Output Fields

Field	Type	Description
content	string
model	string

Examples

Describe image with LLaVA

// Input
{
  "host": "http://localhost:11434",
  "model": "llava",
  "prompt": "What objects are visible in this image?",
  "image_path": "/images/office.jpg"
}

// Output
{
  "content": "I can see a desk with a laptop, a coffee mug, some books...",
  "model": "llava"
}

Install & Discovery

Install

ppm install mcp-vision-ollama

Get JSON Schema

GET /v1/packages/mcp-vision-ollama/1.0.0/schema

Discover by keyword

GET /v1/mcp/discover?q=ollama

Discovery hint: Install with ppm install mcp-vision-ollama or invoke remotely via POST /v1/invoke/mcp-vision-ollama on the MCP Service.

PascalAI Usage

uses toolslib;
var Tool := LoadTool('mcp-vision-ollama');
var R := Tool.Call(JsonObj([]));
Writeln(R.ToJSON);