keiver/image-tiler-mcp-server

Overview

An MCP server that tiles large images and screenshots before sending them to LLM vision models. It prevents downscaling by splitting an image into smaller tiles so that Claude, GPT-4o, and Gemini receive every detail at full resolution.

Best for

Best for
Developers who need to pass detailed large images to LLM vision models without downscaling

Use cases

• Send high-resolution UI screenshots to Claude without losing fine details
• Analyze large diagrams or maps with GPT-4o at original scale
• Provide uncompressed visual context to Gemini for detailed inspection

Notes

An MCP server that tiles large images and screenshots before sending them to LLM vision models. It prevents downscaling by splitting an image into smaller tiles so that Claude, GPT-4o, and Gemini receive every detail at full resolution.

1 stars on GitHub. Last updated 2026-03-07. Licensed MIT.

Use cases

• Send high-resolution UI screenshots to Claude without losing fine details
• Analyze large diagrams or maps with GPT-4o at original scale
• Provide uncompressed visual context to Gemini for detailed inspection

Pros

• Preserves image details that would otherwise be lost due to model resolution limits
• Works with multiple LLMs via the common MCP protocol
• Straightforward integration for developers already using MCP clients

Cons

• Very low popularity (1 star) suggests minimal testing or community validation
• Requires an MCP-compatible client and server setup, adding deployment overhead
• Tiling increases token consumption and processing latency compared to single-image inputs

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.