@tinify-ai/mcp-server

MCP server for Tinify image optimization. AI-powered upscaling, resizing/cropping, compression, and SEO tag generation — all in one tool.

Quick Start

Add to your MCP client config:

{
  "mcpServers": {
    "tinify": {
      "command": "npx",
      "args": ["-y", "@tinify-ai/mcp-server"]
    }
  }
}

No signup required. Works out of the box with 20 free daily credits.

Client-Specific Setup

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "tinify": {
      "command": "npx",
      "args": ["-y", "@tinify-ai/mcp-server"]
    }
  }
}

Claude Code

claude mcp add tinify -- npx -y @tinify-ai/mcp-server

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "tinify": {
      "command": "npx",
      "args": ["-y", "@tinify-ai/mcp-server"]
    }
  }
}

Windsurf

Edit ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "tinify": {
      "command": "npx",
      "args": ["-y", "@tinify-ai/mcp-server"]
    }
  }
}

Tool: `optimize_image`

Optimizes an image with smart defaults: AI-powered upscaling, resizing/cropping, compression, and SEO tag generation — all in one image optimization tool.

Parameters

Parameter	Type	Required	Default	Description
`input`	string	Yes	—	Local file path or remote URL
`output_path`	string	No	auto	Where to save the result
`output_format`	string	No	original	jpeg, png, webp, or original
`output_width_px`	int	No	—	Target width in pixels
`output_height_px`	int	No	—	Target height in pixels
`output_upscale_factor`	float	No	—	Scale factor (2.0, 4.0)
`output_resize_mode`	string	No	pad	pad or crop
`output_aspect_lock`	bool	No	true	Maintain aspect ratio
`output_seo_tag_gen`	bool	No	true	Generate SEO metadata

Examples

Basic optimization:

"Optimize hero.png"

Convert to WebP:

"Optimize hero.png as webp"

Resize and optimize:

"Optimize hero.png to 1920x1080"

Upscale a small image:

"Upscale logo.png by 4x"

Save to specific path:

"Optimize hero.png and save to ./dist/hero.webp"

Batch from URL:

"Optimize https://example.com/photo.jpg"

Output

Returns optimized file path and metadata:

Optimized: hero.tinified.webp
Size: 142.3 KB
Compression: 73%
Format: webp
Dimensions: 1920x1080
Alt text: Modern office workspace with laptop and coffee cup on wooden desk

Structured metadata is also returned with seo_keywords, seo_filename, and full dimension/size data.

Supported Formats

Format	Input	Output
JPEG	Yes	Yes
PNG	Yes	Yes
WebP	Yes	Yes

Max file size: 50 MB.

How It Works

Local file or URL
  → Upload to Tinify API
    → Smart compression (smart lossy, typically 60-80% reduction)
    → AI SEO tag generation (alt text, keywords, filename)
    → Optional: resize, upscale, format conversion
  → Download optimized file
    → Save next to original as name.tinified.ext

All processing happens server-side via the Tinify API. The MCP server is a thin client that orchestrates the pipeline.

Credits

	Free Tier
Credits/day	20
Images/day	5 (with default settings)
Cost per image	4 credits (3 compression + 1 SEO tags)
Signup required	No

Session tokens are stored locally at ~/.tinify/session.json and persist across invocations.

Need more credits? See plans at tinify.ai.

Troubleshooting

Server not appearing in tool list:

Restart your MCP client after editing the config
Ensure Node.js >= 18 is installed: node --version
Try running directly: npx -y @tinify-ai/mcp-server (should start without errors)

"Insufficient credits" error:

Free tier allows 20 credits/day (resets daily)
Each image costs 4 credits with default settings
Disable SEO tags (output_seo_tag_gen: false) to reduce to 3 credits/image

File not found:

Use absolute paths for local files
For URLs, ensure the image is publicly accessible

Timeout errors:

Large images or AI upscaling can take 30-60 seconds
The server has a 60-second timeout per job

Requirements

Node.js >= 18
An MCP-compatible client (Claude Desktop, Claude Code, Cursor, Windsurf, Cline, etc.)

License

MIT - see LICENSE.

@tinify-ai/mcp-server