Krita MCP Server
Allows AI agents to interact with a running Krita instance to create canvases, paint strokes, draw shapes, fill areas, set colors and brushes, export images, and perform document operations such as saving, opening, undoing, and clearing.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Krita MCP v4
This v4 implementation is built from nanayax3/krita-mcp.
This canonical NixOS/Krita 6 bridge is a strict protocol-v4 implementation. It contains:
a PyQt6 Krita plugin with loopback HTTP endpoints;
an asynchronous, typed FastMCP 4.0.3 server;
contract and security tests that run without importing Krita.
There is no v3 or pixel-paint compatibility surface.
Install
Copy krita-plugin/kritamcp/ and krita-plugin/kritamcp.desktop into
Krita's Python plugin directory, enable “Krita MCP Bridge” in the Python Plugin
Manager, and restart Krita. Install the wrapper environment with:
python -m venv .venv
.venv/bin/pip install -r requirements.txtConfigure the MCP client to run this repository's server.py.
Related MCP server: Krita Illustration MCP
Security and transport
The plugin binds only to 127.0.0.1:5678. At startup it creates a random
256-bit bearer token at $XDG_RUNTIME_DIR/krita-mcp/token with mode 0600.
The token is required for /v4/capabilities and /v4/command and is never
logged. /health is the only unauthenticated endpoint.
The bridge validates request size, action names, queue depth, revisions, request-id retries, stroke batch limits, and allowlisted resolved paths. One session owns the writer lease; disconnect or timeout releases it without closing the document.
Protocol
See docs/API.md. The command envelope contains protocol version, request/session IDs, optional document ID, expected revision for every write, an action, and parameters. Mutating replies are cached by request ID within the session, making retries idempotent.
Documents created by v4 are RGBA/U16, 300 PPI, using
sRGB-elle-V2-srgbtrc.icc. KRA save and PNG export are separate operations.
Projection capture explicitly converts to 8-bit sRGB.
Painting accepts one native paintLine per line stroke or one native cubic
paintPath per curved stroke. Every stroke supplies complete brush state.
Line pressure is supported; cubic pressure, tilt, rotation, and speed are
rejected. Candidate transactions use ephemeral paint layers and never snapshot
a full U16 canvas.
Verification
python -m unittest discover -s tests -vHeadless tests cover schemas, authentication, idempotency, revision conflicts, path/symlink traversal, resource identity, unsupported sensors, limits, transaction ownership, and source-level native painting invariants. Live Krita verification requires the user-started application and plugin.
License
MIT
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Generate AI images and videos from any compatible MCP client.
MCP server for OpenAI API (chat completions, image generation, embeddings) via AceDataCloud
Generate AI video, images and audio, and browse your history, from any MCP client.
MCP server for Flux AI image generation
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceEnables AI-driven 2D drawing and animation in Blender, allowing MCP clients to create Grease Pencil drawings, keyframes, materials, and renders directly inside a running Blender instance.MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI models to directly control a local Krita instance by inspecting real canvas previews, managing documents and layers, importing and masking generated images, inpainting, painting with the brush engine, checkpointing, and exporting KRA/PNG files.MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to create, paint, edit, inspect, and export images in a running GIMP 3.2 instance via MCP.178 PyPIGPL 3.0
- AlicenseNot gradedqualityBmaintenanceEnables LLM-driven painting agents to control Krita through MCP tools, iteratively reconstructing a target image by observing canvas snapshots, diff heatmaps, and session state. It exposes observation, drawing, brush, color, layer, selection, vector, filter, document I/O, and session tools for closed-loop artwork generation.MIT