KritaPilot: Krita MCP server and AI assistant

Let AI create and edit art, photos, animation and video inside a running Krita. KritaPilot is a Krita MCP server for any MCP client - Claude, ChatGPT, Gemini, Cursor, VS Code Copilot, Codex and others - and an AI Chat inside Krita that works with 40+ AI providers, including free ones. (Formerly krita-mcp.)

AI client ──MCP (stdio or HTTP)──▶ KritaPilot ──127.0.0.1 + token──▶ Krita plugin ──▶ your canvas

Painted by an AI with Krita's own tools - no image model. The agent laid down procedural skies and water, vector shapes and real brush strokes on separate layers, looked at its work with canvas_snapshot and analyze_image, and improved it. Both are reproducible: examples/galilee_dawn.py and scripts/paint_sunset.py.

Why KritaPilot

• It works where artists work. Everything happens in your open Krita document, on layers you can edit, with undo for every AI step. Nothing is a flat image you have to import.
• Any AI, including free ones. Use it from Claude, ChatGPT, Gemini, Cursor or VS Code over MCP - or chat inside Krita with 40+ providers. Auto switches to the next provider when one runs out of credit, and free models with no account are built in.
• Real tools, not just a prompt box. 86 tools cover Krita's brush engine, layers, selections, filters, animation and video, plus image models (ComfyUI, AUTOMATIC1111, OpenAI, Gemini, ...) when you want them.
• Safe by design. Loopback only, a secret token, one workspace folder, no code execution, keys never shown, read-only and pause switches in the panel.

What the AI can do (86 tools + 7 workflow prompts)

• Paint like an artist. It uses Krita's real brush engine with any preset, pressure and eraser, plus crisp shapes and gradients, editable SVG vectors and text, procedural SeExpr textures, screentones, and colourising line art with Krita's Colourise Mask.
• See and measure. Snapshots (zoomed in, a single layer, any animation frame, or a labelled coordinate grid), plus analyze_image (palette, light, contrast, sharpness, layout), exact read_pixels, and find_color.
• Edit photos. Exact levels, curves, exposure and temperature; hue/saturation, colour balance, sharpen, denoise, blur and about 25 more; auto-enhance; vignette; non-destructive filter masks; 16-bit colour spaces.
• Chat with an AI inside Krita. The AI Chat docker lets you sign in to Gemini, Claude, OpenAI, OpenRouter (with your browser, no key to copy), Mistral, Groq, DeepSeek, xAI Grok, Together, Fireworks, Cerebras, Hugging Face, NVIDIA, Qwen, Moonshot, SambaNova, Cohere, Z.AI, DeepInfra, Vercel AI Gateway and about 20 more, or a local model (Ollama, LM Studio), or use it free with no sign-in (Pollinations, Kilo Gateway, OVHcloud, LLM7, or the OmniRoute gateway), type what you want and watch it work in your document.
  • Auto (the default) uses all your providers in turn - yours first, then the free ones - and when one runs out of credit, hits a quota, rejects its key or stops answering, the next one carries on with the same task.
  • It works like a coding assistant's chat: Enter sends, messages typed while the AI works are queued, Stop stops at once, and a status line shows what it is doing.
  • It keeps spending down (only the newest pictures are resent, Claude's prompt cache is used, repeated failures stop) and shows the tokens each answer used.
  • Skills (39 built in - photoreal images, retouching, pixel art, logos, comics, animation, saving credits, ... - plus any you add) are loaded only when a request needs them.
  • One Services list sets up every service the way it works: an API key, a browser sign-in, a server address with a connection test, or nothing at all for the free ones. Models are listed live from each provider, and AI apps connected through MCP use the same sign-in.
• Real-world materials. CC0 photo textures and HDRI skies from Poly Haven, and openly licensed photos from Openverse (with attribution), tiled or placed on layers to make work look real.
• Just give it a prompt. create_artwork decides: realistic or photographic requests are generated (with looks like photo, cinematic, portrait or product, plus negative prompts that remove AI artefacts). Hand-made requests, or setups with no model connected, are painted in Krita, with prepared layers, guides and a painting plan.
• Generate with any image model, right on the canvas. It works with ComfyUI (including your own workflows: Flux, ControlNet, IP-Adapter, …), AUTOMATIC1111/Forge, OpenAI, Gemini, Stability AI, Replicate and fal.ai. You can generate from text, rework a region (img2img), inpaint a selection, outpaint beyond the edges, upscale, and remove backgrounds. Several variations at once come with a numbered comparison sheet. You can also paint a sketch layer and let ControlNet follow it. match_colors blends a generated element into the scene. Results always land on new layers, cut to your selection, and each placement is one undo step.
• Video and a second opinion. Generate video with Sora, Veo, HeyGen talking avatars (they can even make your painted character talk) or any Replicate/fal video model. It opens in Krita as an animation you can edit. Cinematic camera moves (dolly, orbit, crane, FPV, crash zoom…) steer video models, and camera_move animates any still inside Krita without AI. ask_vision gets a critique, caption or prompt from Claude, GPT or Gemini (or a local Ollama model).
• Animate and edit video. Keyframes, onion skin, frame contact sheets, importing video from mp4/mov/webm/gif, grading every frame (for_each_frame), and rendering to mp4/webm/gif/ProRes with Krita's bundled ffmpeg.
• Full layer control. Paint, vector, group, fill, filter, clone and file layers; masks; clipping; blend modes; arranging; merging; transforms. About 70 of Krita's own commands via krita_action.
• Real undo for AI edits, even for edits Krita itself doesn't record. It never overwrites your own work.
• batch and for_each_frame let the AI do many steps in one call.

Built-in prompts (create, generative_inpaint, concept_variations, retouch_photo, product_shot, paint_scene, animate) appear as slash commands in clients that support MCP prompts.

The plugin updates itself from this project's GitHub releases: it tells you when a new version is out (or installs it automatically, if you choose), checks every download against GitHub's published checksum, and updates the KritaPilot server too.

An AI Bridge (MCP) panel inside Krita shows the status and a live activity log. From it you can make the AI read-only or pause it, undo AI edits, set the workspace and token, and copy a ready-made config for your AI client. Popups Krita opens while the AI works (such as "Missing Colour Profile") are cancelled and reported to the AI instead of blocking it. Logs… follows every command, chat step, setting change and error live, dated to the millisecond and with keys masked. If Krita ever freezes, the plugin records where in ~/.kritapilot/hang.log.

Quick start

1. Plugin. Download kritapilot-plugin-*.zip from Releases. In Krita choose Tools ▸ Scripts ▸ Import Python Plugin from File…, enable the plugin and restart Krita.
2. Server. Install it from PyPI or straight from GitHub:

   pip install kritapilot                                              # PyPI
   pip install "git+https://github.com/jowinjohnchemban/kritapilot"     # GitHub (latest main)
   

   pipx install kritapilot and uvx kritapilot work too, and every release has the wheel. (Not pip install krita-mcp: that is a different project.)
3. Use it, either way:
   • Inside Krita, free: open Settings ▸ Dockers ▸ AI Chat, keep the provider on Auto and type. Free models need no account; sign in to any provider in Services… for faster, stronger results.
   • From your AI app: open Settings ▸ Dockers ▸ AI Bridge (MCP), pick your client and copy the snippet. For example, in Claude Code: claude mcp add krita -- kritapilot
4. Ask: "Create a 1200×800 canvas, paint a sunset over mountains with real brushes, look at it and improve it."

Details: installation · clients · AI Chat · the panel

Command-line flags

kritapilot [--transport {stdio,http}] [--host HOST] [--port PORT] [--agent [--agent-port N]]
          [--check-update | --update] [--version]

Environment variables:

Generative services are configured with more server variables (KRITAPILOT_COMFYUI_URL, KRITAPILOT_A1111_URL, KRITAPILOT_OPENAI_API_KEY, KRITAPILOT_GEMINI_API_KEY, KRITAPILOT_STABILITY_API_KEY, KRITAPILOT_REPLICATE_API_TOKEN, KRITAPILOT_FAL_KEY, KRITAPILOT_HEYGEN_API_KEY, KRITAPILOT_ANTHROPIC_API_KEY, KRITAPILOT_IMAGE_BACKEND, …). Cloud services are used only with a KRITAPILOT_… key, so a key that happens to be in your environment never uploads your art by surprise. See Connecting AI services.

Documentation

Safety in one paragraph

The bridge listens on loopback only and requires a secret token. The AI can read and write files only in one workspace folder (~/KritaMCP by default). There is no code-execution tool. You can switch the AI to read-only or pause it from the panel at any time. Full details: security model.

Status

The project is tested against Krita 5.3.4 on Windows. scripts/smoke_krita.py runs 56 end-to-end checks against a running Krita, including an undo matrix over every kind of edit and the Quick start's sunset prompt. scripts/paint_sunset.py paints the sunset with real brushes, looks at it, improves it and checks the result. The AI Chat has been run end to end against real Krita on Auto with only the free services (Pollinations, Kilo, AI Horde): a photoreal prompt was generated, compared and saved, with the switch to the next provider happening on the way. Unit tests run on Linux and Windows with PyQt5 and PyQt6 in CI. Krita 6 is covered by the PyQt6 unit tests but has not yet been run end to end.

FAQ

Do I need a paid AI account? No. The AI Chat inside Krita works with free models and no account (Pollinations, Kilo Gateway, OVHcloud, LLM7), and AI Horde makes free pictures. They are slower and simpler than paid models; one paid provider (or a local model with Ollama) makes a big difference.

Do I need a GPU? No. Painting with Krita's brushes needs none, and cloud image models run elsewhere. If you have one, ComfyUI or AUTOMATIC1111 run image models locally for free.

Does it upload my artwork? Only to the services you set up, and only when a request uses them (for example, an image model reworking a region, or a snapshot shown to a chat model). Painting, editing and measuring happen in Krita on your computer. Cloud services are used only with keys you saved yourself.

Can the AI damage my files? It can read and write only inside one workspace folder, every edit is a separate undo step, it never overwrites your own layers unless asked, and you can switch it to read-only or pause it at any time.

Which Krita versions? Krita 5.3 is tested end to end on Windows; Linux and macOS use the same plugin. Krita 6 (PyQt6) is covered by the unit tests.

Where do I start? Follow the Quick start, then try the workflows. Something wrong? Open Logs… in the panel; it usually says why.

Contributing

Bug reports, ideas, new skills and pull requests are welcome - see CONTRIBUTING.md. What changed recently is in the changelog. Security problems: SECURITY.md.

pip install -e .[test] PyQt6
QT_QPA_PLATFORM=offscreen pytest
python scripts/smoke_krita.py        # against a real running Krita with the plugin enabled

See development.