Log inSubmit server

Hidden while we check this listing

Our automated checks found a problem, so this listing is out of search, browse and the API until it clears. It comes back on its own once the problem is fixed.

  • Package no longer on the registry
Own this project? Claim it
MCP Florence2 logo

MCP Florence2

Runs Florence-2 image OCR, captioning, and custom-prompt processing through MCP.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install MCP Florence2

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "mcp-florence2": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/jkawamoto/mcp-florence2",
        "mcp-florence2"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

Image OCR, captioning, and prompt-based processing are exposed by the mcp-florence2 MCP server through Florence-2. It accepts a local file path or URL as the image source, with custom prompts available through the process tool. Use it when an MCP client needs to extract text from images or PDFs or generate image descriptions. The server uses stdio configuration and can be installed in Claude Desktop, Goose, or LM Studio.

Use cases

  • Extract text from scanned images
  • Generate captions for visual files
  • Analyze an image with a custom prompt
  • Process locally stored or web-hosted files

Key features

  • Florence-2 image processing
  • OCR text extraction
  • Image caption generation
  • Custom prompt processing
  • Local path and URL inputs
  • Stdio MCP configuration

Tools2

What MCP Florence2 lets your agent do. Expand a tool for its parameters.

Self-reported~29 tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 3d ago

Recent health check succeeded.

5/5 checks healthy over the last 41d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 14d ago via OSV.dev · mcp-florence2 (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars7Stargazers on the source repository.
Last commit22d agoMost recent push to the default branch.
Tools exposed2Callable tools this server registers over MCP.

How MCP Florence2 works

What the mcp-florence2 MCP server does

The mcp-florence2 MCP server connects MCP clients to Microsoft’s Florence-2 model for image understanding tasks. It supports extracting visible text with OCR, producing descriptive captions, and sending a custom instruction to the model for image processing. Sources are provided as local file paths or URLs. The README also identifies images and PDF files as supported processing inputs, although the tool argument is described as an image file path or URL.

This server fits workflows where an agent needs to inspect visual files without a separate image-analysis application. Typical inputs include scanned documents, screenshots, photographs, and other files that can be referenced locally or from a web server.

How it works

Each tool receives a src value identifying the file to process. That value may point to a local file or a URL. The selected operation then uses Florence-2 for the requested task:

  • OCR returns text recognized in the source.
  • Captioning creates a description of the image content.
  • Custom processing accepts a prompt in addition to src, allowing the caller to specify what it wants Florence-2 to analyze or produce.

The server is configured as a stdio MCP extension. The documented command uses uvx to run the package directly from the Git repository, so the MCP client launches the server as a local process.

Setup and configuration

Claude Desktop users can install the released .mcpb bundle or add a manual mcpServers entry. The manual configuration invokes uvx with the repository URL and the mcp-florence2 command, followed by an application restart.

Goose supports installation through its extension link or through a YAML extension entry. That entry uses the same uvx command and declares the extension type as stdio. LM Studio provides a dedicated installation link containing the server configuration.

No environment variables are documented as required. The repository states that the application is licensed under MIT.

Tools and capabilities

The mcp-florence2 MCP server exposes three tools:

  • ocr: accepts src and extracts text from the referenced file with optical character recognition.
  • caption: accepts src and generates captions describing the image.
  • process: accepts src and prompt, then processes the file according to the custom Florence-2 instruction.

Use ocr when the required output is text, caption when a general description is sufficient, and process when the agent needs to provide a task-specific instruction.

Limitations and notes

It also documents setup instructions for Claude Desktop, Goose, and LM Studio only. Inputs must be addressable by a local path or URL, and custom instructions are available through process rather than the OCR or caption tools.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about MCP Florence2

mcp-florence2 is an MCP server that connects MCP clients to the Florence-2 model for image processing. Its main tools are `ocr` for text extraction, `caption` for image descriptions, and `process` for custom prompts. It accepts local file paths or URLs as sources.

Install it through the Claude Desktop bundle, Goose extension link, or LM Studio installation link. Manual setups run `uvx` with the Git repository URL and the `mcp-florence2` command as a stdio server.

Use the `ocr` tool with `src` set to a local file path or URL. The tool sends the referenced image for optical character recognition and returns the extracted text.

Use the `process` tool with both `src` and `prompt`. `src` identifies the local or web-hosted image, while `prompt` tells Florence-2 what processing task to perform.

Related MCP servers

All in Multimedia Process All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewMCP Florence2 AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/mcp-florence2?style=directory)](https://allmcps.com/mcp/mcp-florence2)
HTML Embed
<a href="https://allmcps.com/mcp/mcp-florence2"><img src="https://allmcps.com/api/badge/mcp-florence2?style=directory" alt="MCP Florence2 on AllMCPs" /></a>