Skip to main content

Start from Prompt

With Código, you can describe what you want to build in plain language, and DevAI will handle the rest.
From a simple idea to a full program structure, prompts are the fastest way to kickstart your smart contract project.


1️⃣ Create Your Workspace

Start by creating a new workspace in Código Hub.
Click the green NEW button in the top-left corner of the dashboard.

You’ll be prompted to enter:

  • A workspace name
  • A smart contract name
Workspace creation

2️⃣ Prompt Your Need

Once inside your workspace, tell the AI what you want to build.

You can keep it simple:

generate a counter solana program with create_counter and increment instructions

Or be more specific and describe:

  • The list of instructions
  • The on-chain data you want to store
  • Business rules or logic you want enforced
Prompt counter

3️⃣ AI Conception

The AI will respond with a detailed plan:

  • Overview of what it's going to build
  • Instructions and their roles
  • Accounts & PDAs
  • Custom errors
  • State structure and types

If something doesn’t look right, feel free to iterate:

  • Ask follow-up questions
  • Add or modify logic
  • Change names, flows, or rules

Once you're satisfied, just reply and choose your framework: Anchor or Native Rust. :

yes, with anchor

4️⃣ Wait for Generation to Complete

Código will generate the full program based on your prompt and selected framework.
The workspace will be populated with all source files, pre-wired tests, and boilerplate.

counter

5️⃣ Continue Building with AI

Your project is ready, but the business logic inside the instructions is left empty on purpose.

It's now your turn to guide the AI to implement each one, using natural prompts like:

Implement the file initialize_counter with basic verification

You can iterate, review, and apply changes with full control.

counter

Repeat this process for each instruction file.


🚀 Next Steps

👉 How to Build & Deploy My Program
👉 How to Write and Run Unit Tests
👉 How to Implement Business Logic with AI Autocompletion