MCP Tool Reference

The floimg MCP server exposes these tools for AI agents.

floimg_generate

Generate an image from parameters.

Parameters

{
  generator: string;    // Generator name
  params: object;       // Generator-specific parameters
  output?: string;      // Output path (optional)
}

Example

{
  "generator": "quickchart",
  "params": {
    "type": "bar",
    "data": {
      "labels": ["A", "B", "C"],
      "datasets": [{ "data": [10, 20, 30] }]
    }
  },
  "output": "chart.png"
}

Available Generators

Generator	Description
quickchart	Chart.js charts
mermaid	Diagrams and flowcharts
qr	QR codes
d3	D3 visualizations
screenshot	Web page screenshots
gemini-generate	AI image generation (requires @teamflojo/floimg-google)
google-imagen	Google Imagen models (requires @teamflojo/floimg-google)

---

floimg_transform

Transform an existing image.

Parameters

{
  input: string;        // Input image path or URL
  op: string;           // Transform operation
  params: object;       // Operation parameters
  output?: string;      // Output path (optional)
}

Example

{
  "input": "chart.png",
  "op": "resize",
  "params": {
    "width": 800,
    "height": 600
  },
  "output": "chart-resized.png"
}

Available Operations

Operation	Parameters	Description
resize	width, height, fit	Change dimensions
blur	sigma	Apply gaussian blur
sharpen	sigma	Sharpen edges
rotate	angle	Rotate by degrees
crop	left, top, width, height	Crop to region
convert	format, quality	Change format
edit	prompt, provider, apiKey	AI image editing (see below)

AI Transform (Edit)

Use the edit operation with an AI provider:

{
  "input": "photo.jpg",
  "op": "edit",
  "params": {
    "provider": "gemini-transform",
    "prompt": "Make the sky more vibrant with sunset colors",
    "apiKey": "your-google-ai-key"
  },
  "output": "enhanced-photo.jpg"
}

Available AI transform providers:

• gemini-transform - Gemini image editing (requires @teamflojo/floimg-google)

---

floimg_save

Save an image to storage.

Parameters

{
  image: string;        // Image reference from generate/transform
  path: string;         // Destination path
}

Example

{
  "image": "generated-chart",
  "path": "s3://my-bucket/charts/report.png"
}

Supported Destinations

Prefix	Description
./ or /	Local filesystem
s3://	Amazon S3
custom://	Registered custom storage

---

analyze_image

Analyze an image using AI vision.

Parameters

{
  image: string;        // Image path or URL
  prompt?: string;      // Analysis prompt (default: "Describe this image")
  provider?: string;    // Vision provider (default: auto-detected)
  apiKey?: string;      // API key for the provider
  outputFormat?: 'text' | 'json';  // Response format
  jsonSchema?: object;  // Schema for structured output
}

Example

{
  "image": "product.jpg",
  "prompt": "List all products with their colors and estimated prices",
  "provider": "gemini-vision",
  "outputFormat": "json",
  "jsonSchema": {
    "type": "object",
    "properties": {
      "products": {
        "type": "array",
        "items": {
          "type": "object",
          "properties": {
            "name": { "type": "string" },
            "color": { "type": "string" },
            "price": { "type": "string" }
          }
        }
      }
    }
  }
}

Available Providers

Provider	Description
gemini-vision	Gemini vision (requires @teamflojo/floimg-google)
grok-vision	Grok vision (requires @teamflojo/floimg-xai)

---

generate_text

Generate text using AI.

Parameters

{
  prompt: string;       // Text prompt
  provider?: string;    // Text provider (default: auto-detected)
  apiKey?: string;      // API key for the provider
  context?: string;     // Additional context
  systemPrompt?: string; // System instructions
  outputFormat?: 'text' | 'json';  // Response format
  jsonSchema?: object;  // Schema for structured output
}

Example

{
  "prompt": "Generate 3 creative prompts for product photography",
  "provider": "gemini-text",
  "outputFormat": "json",
  "jsonSchema": {
    "type": "object",
    "properties": {
      "prompts": {
        "type": "array",
        "items": { "type": "string" }
      },
      "style": { "type": "string" }
    }
  }
}

Available Providers

Provider	Description
gemini-text	Gemini text (requires @teamflojo/floimg-google)
grok-text	Grok text (requires @teamflojo/floimg-xai)

---

run_pipeline

Execute a multi-step workflow.

Parameters

{
  steps: Array<{
    type: 'generate' | 'transform' | 'text' | 'vision';
    params: object;
  }>;
  output?: string;      // Final output path
}

Example

{
  "steps": [
    {
      "type": "vision",
      "params": {
        "image": "input.jpg",
        "prompt": "Describe the mood of this image",
        "provider": "gemini-vision"
      }
    },
    {
      "type": "text",
      "params": {
        "prompt": "Based on this mood description, suggest an edit",
        "context": "{{previous.content}}"
      }
    },
    {
      "type": "transform",
      "params": {
        "input": "input.jpg",
        "op": "edit",
        "provider": "gemini-transform",
        "prompt": "{{previous.content}}"
      }
    }
  ],
  "output": "enhanced.jpg"
}

---

Provider Auto-Discovery

When @teamflojo/floimg-google or @teamflojo/floimg-xai are installed, the MCP server automatically discovers and registers their providers:

floimg-google provides:

• gemini-text - Text generation
• gemini-vision - Image analysis
• gemini-transform - Image editing
• gemini-generate - Image generation

floimg-xai provides:

• grok-text - Text generation
• grok-vision - Image analysis

To use these providers, install the package:

npm install @teamflojo/floimg-google
# or
npm install @teamflojo/floimg-xai

Then restart the MCP server. The providers will be automatically available.

---

Tool Responses

All tools return a response object:

Success

{
  "success": true,
  "path": "/path/to/output.png",
  "width": 800,
  "height": 600,
  "format": "png",
  "size": 45678
}

Error

{
  "success": false,
  "error": "Generator not found: unknown"
}

---

Chaining Tools

AI agents can chain tools together:

1. Generate a chart
2. Transform to resize and convert
3. Save to S3

The MCP server tracks image references between calls, so subsequent operations can reference previous outputs.

See Also

• MCP Overview - Introduction to MCP
• Claude Integration - Setup guide
• Examples - Common use cases