Works with
Anthropic ClaudeOpenAIGoogle GeminiOllamaMCP servers
How it works
Three steps from idea to schematic
Knit closes the loop between natural language, code, and a real KiCad schematic.
01
Describe the circuit
Tell Knit what you want to build in plain language — a power supply, a microcontroller breakout, a sensor front-end.
02
Knit generates SkiDL
The AI writes SkiDL (Python) describing your nets and parts, then runs it in a sandbox to validate the netlist.
03
Preview & export to KiCad
Get a schematic preview and a KiCad export. Refine in chat or edit the code by hand — both sides stay in sync.
Features
Built for how engineers actually work
Own your keys, own your code, own your export.
Credits, not subscriptions
Buy any amount you like, from €1 to €100. Credits never expire — no monthly lock-in.
Bring your own API key
Use your own LLM key (Anthropic, OpenAI, Gemini, or local Ollama). BYO runs consume zero Knit credits.
Editor + chat, same files
Edit the SkiDL code by hand and collaborate with the AI on the exact same project files.
Sandboxed execution
Generated SkiDL runs in an isolated sandbox, so the netlist is validated before you ever see a schematic.
KiCad export
Export your design to KiCad and continue in your existing EDA workflow — no lock-in to a proprietary format.
MCP tools
Connect Model Context Protocol servers to extend the AI with your own tools and data sources.
Integrations
Connect your AI and your tools
Bring your own provider and your own MCP servers. Each guide walks you through it.
Anthropic Claude
Bring your own Anthropic key and generate schematics with Claude.
How to connect →OpenAI
Use OpenAI — or any OpenAI-compatible gateway via a custom base URL.
How to connect →Google Gemini
Connect Gemini with just your API key through the official endpoint.
How to connect →Ollama
Run open models locally (via a public tunnel) or on Ollama Cloud.
How to connect →MCP servers
Add Streamable-HTTP MCP servers to give the AI extra tools.
How to connect →FAQ
Frequently asked questions
- What is Knit?
- Knit is an AI tool that turns a plain-text description of a circuit into a KiCad schematic. You describe the circuit in chat; Knit generates SkiDL (Python), runs it in a sandbox, and returns a schematic preview and a KiCad export.
- Which AI providers can I use with Knit?
- Knit supports Anthropic (Claude), OpenAI (and OpenAI-compatible endpoints), Google Gemini, and Ollama (local or cloud). You can bring your own API key for any of them.
- How much does Knit cost?
- Knit uses pay-as-you-go credits — buy any amount from €1 to €100, and they never expire. If you bring your own LLM API key, those runs consume zero Knit credits.
- Does bringing my own API key cost Knit credits?
- No. When you use your own LLM key, Knit calls the model with your credentials and the run costs zero Knit credits. You pay your AI provider directly for the tokens.
- Can I edit the generated code?
- Yes. Knit puts a code editor and the AI chat side by side on the same project files, so you can edit the SkiDL by hand and keep collaborating with the AI.
- Can I extend Knit with my own tools?
- Yes. You can connect Model Context Protocol (MCP) servers over Streamable HTTP. When enabled, their tools are offered to the AI during chat.