image-mcp

image-mcp

中文 | English

A local, Pillow-based image-processing MCP server (offline, no API key). Via MCP it exposes a set of image tools to DeepSeek Harness (or any stdio-MCP client): info / resize / crop / convert / compress / rotate / flip / thumbnail / watermark / effects / placeholder / overlay — 12 tools, model-visible as mcp__images__*.

Features

• Offline & zero-cost: local Pillow processing only, no online API or key required.
• 12 image tools: info / resize / crop / convert / compress / rotate / flip / thumbnail / watermark / effects / placeholder / overlay.
• Standard protocol: JSON-RPC 2.0, newline-delimited stdio MCP; all logs go to stderr (stdout is reserved for the protocol).
• Easy wiring: one config entry via @deepseek-ai/dsh-mcp-client in a DSH profile.

Install & wire in (DSH)

pip install -r requirements.txt   # or: pip install Pillow

In a DSH profile patch (e.g. cordis.patch.yml):

- id: mcp-images
  name: '@deepseek-ai/dsh-mcp-client'
  config:
    serverName: images
    transport: stdio
    command: python3
    args: ['<PATH-TO>/image_mcp.py']

Restart DSH and the agent can call the mcp__images__* tools.

Run standalone (no DSH)

echo '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}' | python3 image_mcp.py
# returns the schema of the 12 tools

Tools

Protocol

• Transport: stdio, one JSON object per line.
• Version: JSON-RPC 2.0, Protocol 2025-03-26.
• SERVER_NAME=dsh-image-mcp, SERVER_VERSION=0.1.0.

Supported platforms

• Supported platforms: macOS / Windows / Linux (Python 3 + Pillow, cross-platform).
• Windows note: use python instead of python3 in the MCP config; use absolute or backslash paths in args.

License

MIT © TOBYCAI