Code with AI

Code with AI writes, audits, and deploys contracts for you \u2014 in the right language for the network you\u2019re on. It has two modes: a Chat that drafts and reviews contracts, and an autonomous Agent that builds and ships one end to end.

Chat mode

Describe what you want and the assistant writes production-grade Solidity, or paste an existing contract and ask for a security review (findings are graded by severity). Every generated code block has an Open in Editor button that drops the source straight into the Contract Editor to compile and deploy. The assistant tailors its language and idioms to the network you have selected.

Agent mode

Switch to the Agent tab and describe a contract. The agent then runs the whole build loop on its own, reporting each step inline:

  1. 1
    Generate

    Writes a complete, secure contract from your request.

  2. 2
    Compile

    Compiles in the browser and reads the solc output.

  3. 3
    Auto-fix

    If compilation fails, it rewrites the source and recompiles, up to 5 attempts.

  4. 4
    Deploy

    Deploys with your connected wallet. When the constructor needs arguments, a form appears pre-filled with sensible values for you to review and confirm before signing.

  5. 5
    Record

    Saves the deployment to My Projects and links to the built-in explorer.

Agent deploys are real, wallet-signed transactions on the network you have selected — whichever that is. You approve every transaction in your wallet, where the actual cost is shown before you sign, and the run is saved so it survives a page refresh.

Providers

Both modes run against your chosen AI provider. The assistant can use a server-side proxy so your provider key never reaches the browser, or a direct bring-your-own-key path stored only in your browser.

  • OpenRouter (default) — one key reaches all 15 models below
  • Anthropic: Claude Opus 5, Sonnet 5, Fable 5, Opus 4.8, Haiku 4.5
  • OpenAI: GPT-5.6 Sol / Terra / Luna, GPT-5.5, GPT-5.4 Mini
  • DeepSeek: V4 Pro, V4 Flash
  • Google Gemini 3.7 Flash, xAI Grok 4.6, Qwen3.8 Max
  • Or go direct to OpenAI / Anthropic with that vendor's own key

Pick a provider, model, and key in Settings. The model dropdown is grouped by vendor and shows each model's indicative price per 1M tokens, so the cost tradeoff is visible before you pick.

DevStation
Loading console…