PicEte MCP Integration Guide

Use PicEte's image processing tools from Claude Code, Cursor, Claude Desktop, or any MCP-compatible AI client.

What is MCP?

Model Context Protocol (MCP) is an open standard that lets AI agents (like Claude Code, Cursor) call external tools directly. With picete-mcp, your AI can process images — compress, convert, resize, extract colors — without leaving your terminal and without uploading to any server.

All processing runs locally via Sharp. Zero data leaves your machine. Same privacy guarantee as PicEte.

Quick Start

Install and run with a single command (no signup, no config needed for testing):

npx -y picete-mcp

npm Package: picete-mcp@1.0.0

Platform Configuration

Add the following to your AI client's MCP configuration:

Claude Desktop

Edit claude_desktop_config.json:

{
  "mcpServers": {
    "picete": {
      "command": "npx",
      "args": ["-y", "picete-mcp"]
    }
  }
}

Cursor

Edit .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "picete": {
      "command": "npx",
      "args": ["-y", "picete-mcp"]
    }
  }
}

Claude Code

Edit ~/.claude/settings.json:

{
  "mcpServers": {
    "picete": {
      "command": "npx",
      "args": ["-y", "picete-mcp"]
    }
  }
}

Available Tools

picete-mcp provides 13 tools across two categories:

Atomic Tools

Tool	Function	Key Parameters
`convert`	Format conversion	source, format (jpeg/png/webp/gif), quality
`resize`	Resize image	source, width, height, fit, keep_aspect_ratio
`compress`	Compress file size	source, quality, target_format, max_size_bytes
`extract-colors`	Extract dominant colors	source, color_count (default 5)
`image-to-base64`	Convert to base64/data URL	source, format, data_url
`split-image`	Split into grid tiles	source, rows, cols, overlap_px
`metadata`	Get image metadata	source
`batch`	Chain multiple operations	sources[], operations[]

Smart Tools

Tool	Function	Key Parameters
`optimize-for-web`	One-click web optimization	source, max_width, quality, output_format
`prepare-for-vision-api`	AI Vision preprocessing + token estimation	source, max_longest_side, quality
`favicon`	Generate favicon suite (ico + png)	source, output_dir
`compare-images`	Compare two images (MSE/SSIM/diff)	source1, source2, metric
`collage`	Stitch images together	sources[], direction, gap

Usage Examples

Ask your AI in natural language — it will call the right tool automatically:

You Say	Tool Called
"Compress this image to under 200KB"	compress
"Convert this PNG to WebP"	convert
"Resize this photo to 1200px wide"	resize
"Generate all favicon sizes from this logo"	favicon
"What colors are in this image?"	extract-colors
"Optimize this for my website"	optimize-for-web
"Prepare this image for GPT-4 Vision"	prepare-for-vision-api
"Split this image into a 3x3 grid"	split-image
"Compare these two screenshots"	compare-images
"Get the metadata of this image"	metadata
"Stitch these images horizontally"	collage
"Convert this to base64"	image-to-base64
"Batch resize and compress these 10 images"	batch

Development

# Clone and build
git clone https://github.com/0000wei/picete-mcp.git
cd picete-mcp
npm install
npm run build

# Start the MCP server
npm start

← Back to PicEte