Skip to main content

Choose Your Tools

You put three things in place before writing code: a library (SDK) that builds transactions in your language, a provider that connects it to the chain, and current Cardano context for your AI assistant, if you code with one.

None of these lock you in. Every SDK produces transactions against the same ledger rules, so what you learn about UTXOs, datums, and fees carries over if you switch. Providers sit behind one interface in most SDKs, so changing one is a config line.

Start from a template, or from scratch

If you would rather have something running before you understand it, start from a template. The templates gallery has wallet-connected starters that already wire an SDK, a provider, and a frontend framework together, each with the command that copies it into a new project.

The rest of this page is the setup those templates did for you. Skim it now and come back when you want to change one of the pieces.

Your language, and its SDK

An SDK handles the parts of Cardano you do not want to reimplement: assembling and balancing a transaction, coin selection, fee calculation, CBOR serialization, key derivation, and talking to a provider. You work in your own language and it produces bytes a node accepts.

The code tabs across this curriculum use Evolution and Mesh, both TypeScript. That is the only reason they appear here: with one of them installed, every example on the site runs as written. Nothing you learn depends on either one.

Cardano has SDKs in Python (PyCardano), Rust (Whisky), Go (Apollo), C# (Chrysalis), Java, Swift, and more, alongside lower-level serialization libraries. Builder Tools lists them by language and by what each one covers.

Install it

npm install @evolution-sdk/evolution

Anything else installs with its own package manager. Its Builder Tools entry links to the repository, where the install line lives.

How your code reaches the chain

Your SDK does not reach the chain on its own. It sits on top of a provider, which runs the node infrastructure and exposes the chain through an API, so you can read UTXOs and submit transactions without operating a node yourself. The full path is your code → SDK → provider → node → chain.

Query the chain compares the providers on hosting, keys, and rate limits, and shows how to configure one on a client. Connecting to the chain maps the full range, from a hosted API to running the infrastructure yourself.

To follow along you need a key, and Blockfrost has a free tier:

  1. Sign up at blockfrost.io.

  2. Create a project and select Preprod, the network these examples use.

  3. Copy the project ID. That is your API key, and it starts with preprod.

  4. Store it in an env var. Never commit it, and never ship it in client-side code:

    # .env
    BLOCKFROST_API_KEY=preprodYourProjectIdHere

Context for your AI assistant

Model training data on Cardano drifts. APIs change, libraries get renamed, and patterns move faster than models are retrained, so an assistant left to its training data writes code against a version of the ecosystem that no longer exists.

Cardano Dev Skills is the Cardano Foundation's answer: Markdown skills and bundled documentation refreshed weekly from upstream repositories, usable by any agent that reads Markdown. In Claude Code:

/plugin marketplace add cardano-foundation/cardano-dev-skills
/plugin install cardano-dev-skills@cardano-dev-skills

Then run /cardano-context once per project. Set up your AI assistant covers other agents, what the skills contain, and the extra context individual SDKs ship on top.

Beyond writing code

An assistant can also read live chain state and draft transactions for you to sign. Connect an AI assistant with MCP covers that.

What you are not choosing yet

  • The language you write smart contracts in. A separate toolchain, covered in Smart contracts. Every smart contract example in this curriculum is Aiken.
  • Where you run the chain. Preprod is enough to start. You can also run a chain yourself, with block times you set, on the next page.

Next steps