mcp-florence2 — Claude Code MCP server
This MCP server connects Florence-2 to an agent for OCR, image captions, and custom image processing. It can work with local files and URLs.
Open on GitHubWhat mcp-florence2 does
mcp-florence2 sends images and PDF files to Florence-2 for analysis. Its ocr tool pulls text from an image, caption creates a description, and process accepts a custom prompt alongside the source image.
Each tool takes src, which can be a file path or URL. process also needs prompt, so the agent can request a focused result from the model. The manual setup runs the server through uvx from jkawamoto/mcp-florence2; that launcher is required for this configuration to start.
What it helps with
- extract text from an image with OCR
- generate a caption for a screenshot
- send an image to Florence-2 with a custom prompt
- read content from a local file or image URL
Example prompts for the agent
- Read the text in the image at
srcand return the OCR result. - Create a concise caption for the image at
src. - Process
srcwith this prompt: identify every form field visible in the image.
How to install mcp-florence2
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about mcp-florence2
How do I install mcp-florence2 in Claude Code?
Add an MCP server that runs uvx --from git+https://github.com/jkawamoto/mcp-florence2 mcp-florence2, then restart the client if it does not load the new configuration.
What does mcp-florence2 do?
It connects Florence-2 to an MCP client for OCR, image descriptions, and prompt-based image processing. The source can be a file path or a URL.
Which tools does mcp-florence2 provide?
ocr extracts text, caption writes an image description, and process sends an image plus a custom prompt to Florence-2.
