Start free

Build your backend with Claude Code.

Claude Code writes your backend in TypeScript with the Xano SDK, you run it on the Xano Engine on your machine, then deploy it to Xano. One set of rules covers every agent, and Xano runs the path to production without the DevOps overhead.

Build with Claude Code + Xano How coding agents build on Xano
$ npx @xano/sdk agent-skill install
Coding agent Claude Code writes the TypeScript + Platform Xano runs it in production xano-backend agent skill init AGENTS.md Xano Engine your machine Xano cloud live URL local MCP server one set of rules · one path to production

Built for standardized agent output

The Xano SDK is a deterministic, fully typed authoring surface. Claude Code emits TypeScript that always compiles to a valid, importable workspace, grounded in llms.txt and the AGENTS.md in your project.

Designed to be agent-safe

The Xano Engine runs your backend on your machine with no account and no network, so Claude Code iterates against a real backend without credentials. A disposable ephemeral tests the same code on Xano’s cloud.

Ready for production

Claude Code can generate your app in minutes. With a clear picture of what got built and enterprise-grade infrastructure, Xano makes it production-ready.

Four stops on the SDK line

How to build backends with Claude Code.

  1. 01 · Skill

    Install the agent skill.

    In Claude Code, run /plugin marketplace add xano-sdk/sdk, then /plugin install xano@xano-sdk. Or install the xano-backend skill into every coding agent on your machine at once. When a project needs a backend and has none, the skill has Claude Code confirm Xano with you, run init, deploy, and follow the AGENTS.md that init wrote.

    $ npx @xano/sdk agent-skill install
    Claude Code > /plugin marketplace add xano-sdk/sdk > /plugin install xano@xano-sdk ✓ xano-backend installed # or: npx @xano/sdk agent-skill install SKILL.md xano-backend Claude Code skill ready confirm Xano · run init · deploy · follow AGENTS.md
  2. 02 · Init

    Scaffold and connect.

    init writes a TypeScript backend under xano/, a React or Svelte frontend, the xano:deploy scripts and an AGENTS.md brief. It also connects Claude Code to the Xano Engine’s local MCP server, where Claude Code reads tables and rows, seeds data, and runs functions and tests. Code still changes in xano/ and ships with a deploy. Already on Xano? Add --from workspace to read your workspace back as TypeScript.

    $ npx @xano/sdk init my-app && cd my-app
    my-app/ xano/index.ts TypeScript backend frontend/ React or Svelte AGENTS.md brief for agents package.json xano:* scripts Xano Engine MCP server on your machine Claude Code connected read tables seed data run tests
  3. 03 · Prompt

    Prompt and build.

    Open your project in your IDE with Claude Code running in the integrated terminal. Describe what you need (APIs, workflows, data models) and Claude Code writes the backend in TypeScript with the Xano SDK, grounded in the SDK’s llms.txt and the AGENTS.md in your project, instead of a guess.

    IDE integrated terminal Claude Code Ask APIs Workflows Data models Grounded in llms.txt AGENTS.md ✓ always compiles
  4. 04 · Ship

    Validate and ship.

    npm run xano:deploy typechecks and deploys the backend to the Xano Engine on your machine. Run npx xanosdk test run-all, open the builder link the deploy prints to see what your code compiled to, and redeploy in seconds. When it’s ready to share, npm run xano:deploy:ephemeral puts the same code on a live URL on Xano’s cloud.

    $ npm run xano:deploy
    $ npm run xano:deploy → Deploying ./xano/index.ts → Xano Engine ✓ Xano Engine xanosdk-4f2a91 deployed http://127.0.0.1:53358 Open the builder: http://127.0.0.1:53358/… Xano Engine your machine Ephemeral Xano’s cloud Workspace via a release the same code, from your laptop to production
Set up Claude Code

The skill install, and the AGENTS.md init writes

init writes an AGENTS.md brief, and Claude Code reads it as your project instructions. Install the skill and Claude Code follows the same brief as every other agent on the project.

Install the skill in Claude Code

Run /plugin marketplace add xano-sdk/sdk, then /plugin install xano@xano-sdk, inside Claude Code, and Claude Code has the xano-backend skill.

Or in every coding agent at once

npx @xano/sdk agent-skill install installs the same skill into each coding agent on your machine. status and uninstall work the same way, and npx skills add xano-sdk/sdk is the skills-installer route.

One brief for every agent

Claude Code reads AGENTS.md at the start of every session, alongside the SDK’s llms.txt. Cursor, Codex and the rest read the same file, so every agent on the project builds to the same rules.

Terminal: install the xano-backend skill
# Inside Claude Code
/plugin marketplace add xano-sdk/sdk
/plugin install xano@xano-sdk

# Or into every coding agent on this machine
npx @xano/sdk agent-skill install
npx @xano/sdk agent-skill status
Guide

Make AI-generated code safe for production

AI can generate apps fast. Xano can help you make them production-ready, reliable, and built to last.

How Xano governs AI-generated code
From your agent to production Claude Code writes the code Types compiles Tests suite passes Release read back into production only when every check passes
Docs

Go deeper on the SDK

Every command on this page comes from the Xano SDK. See how coding agents build on Xano →

SDK

The Xano SDK

Guide

Coverage & agent grounding

GitHub

The Xano SDK on GitHub

The platform under the line

Everything Claude Code needs to ship real backends.

Logic in Canvas View

See what your code compiled to as a Function Stack in Canvas View: transparent, auditable, human-readable.

Managed Postgres

Production-ready PostgreSQL with triggers, relationships, and direct access. No migrations, no DevOps.

API builder

Instant REST APIs with auth, rate limiting, and custom logic. Ready to connect to any frontend or service.

AI-assisted building

Describe your backend. AI builds schemas, APIs, auth, and logic, governed from the start.

Xano SDK + agent skill

Write the backend as typed TypeScript in your repo, from any IDE or agent, with full governance intact.

Observability & monitoring

Request tracing, performance dashboards, and audit logs across every system.

Environment management

Branch, test, and promote. Git-based versioning so nothing untested reaches production.

RBAC & compliance

HIPAA, SOC 2, GDPR, and ISO compliance with fine-grained access control. Built in from day one.
Questions

Claude Code and the Xano SDK

Can Claude Code work on my existing Xano workspace?

Yes. npx @xano/sdk init my-app --from workspace reads your workspace back as readable TypeScript (real s.db.query(...), not a JSON dump) and diffs it against the source. Claude Code then works from your real schemas and endpoints. A pull carries schema only, so try changes on the Xano Engine or an ephemeral first.

What does the agent skill do?

xano-backend is an agent skill (SKILL.md). When a project needs a backend and has none, it has Claude Code confirm Xano with you, run init, deploy, and follow the AGENTS.md that init wrote. Install it in Claude Code with /plugin marketplace add xano-sdk/sdk and /plugin install xano@xano-sdk, or with npx @xano/sdk agent-skill install.

How does Claude Code know the SDK’s API?

The package ships machine-readable grounding generated from the SDK’s own sources, so it cannot drift from them: llms.txt, the router with the mental model and the gotchas, the llms/*.md topic files it maps, and manifest.json. init also drops an AGENTS.md brief into every project.

What does the Xano Engine’s MCP server give Claude Code?

init connects Claude Code to the MCP server of the Xano Engine running on your machine. Through it, Claude Code reads tables and rows, seeds data, and runs functions and tests against a real backend, with no account. Code still changes in xano/ and ships with a deploy. npx xanosdk local mcp shows how an agent connects.

What happens if Claude Code generates something wrong?

The typed surface catches a lot before it runs: a backend rename is a compile error rather than a 404. The tests you author run with npx xanosdk test run-all, and a failing suite exits 5, distinct from a crash. Every change can go to a disposable ephemeral first, and nothing reaches your workspace without a release. You’re never deploying blind.

Can Claude Code share a project with other coding agents?

Yes. Every agent reads the same AGENTS.md that init wrote, so Claude Code and a teammate in Cursor or Codex build to one brief. The xano-backend skill also installs into Copilot, Gemini CLI, Amp and OpenCode. See how coding agents build on Xano.

Do I need a Xano account to start?

No. init needs no sign-in, and the Xano Engine runs your backend on your machine with no account and no network. When you want a URL to share, sign in once with npx xanosdk login and deploy to an ephemeral on Xano’s cloud.

Will a backend Claude Code writes scale?

Yes, because scaling is the platform’s job, not the code’s. Claude Code writes the logic; Xano runs it on managed Postgres with auth, background jobs and deployment handled for you. The same TypeScript runs on the Xano Engine on your machine and, after a release, on Xano’s cloud.

What if I outgrow AI-generated logic?

AI is great at going from zero to one, but sometimes you need more careful control to refine, modify, and extend what was generated. Everything Claude Code writes is TypeScript in your repo: version it, review it in pull requests, edit it by hand, and redeploy. Canvas View shows what it compiled to. You’re never locked into output you can’t understand or change.

AI builds software. Xano makes it trustworthy.

The backend agents can build and humans can trust.

Start free Talk to sales