Describe it. Deploy it.

Tell fold what your tool should do in plain English. AI generates the code, bundles it, and gives you a live endpoint with MCP and A2A support.

Idea to endpoint in one command

No boilerplate. No setup. Describe your tool and let AI handle the implementation.

Run fold create-tool with a name and a plain English description of what your tool should do.

fold generates production-ready TypeScript that uses the runtime SDK. The code handles input parsing, validation, and structured output.

Your tool is bundled, versioned, and pushed to the edge. You get an HTTPS URL, an MCP endpoint, and an A2A agent card.

One command generates and deploys a fully functional MCP tool.

$ fold create-tool summarizer "Summarize any URL into three bullet points"

Generating tool...

Deploying...

Deployed as MCP tool!

URL: https://you.fold.run/summarizer

MCP endpoint: https://you.fold.run/.well-known/mcp

Agent card: https://you.fold.run/.well-known/agent.json

AI-generated code is not a black box. Download it, modify it, and redeploy with fold deploy.

// Generated: summarizer.ts

import { defineHandler } from '@fold-run/runtime';

export default defineHandler(async (fold: FoldContext) => {

const { url } = await fold.body<{ url: string }>();

const page = await fetch(url);

const text = await page.text();

const summary = await fold.ai.run('@fold/llama-3', {

prompt: `Summarize in 3 bullet points:\n${text}`

});

return fold.json({ url, summary });

});

# Edit locally, then redeploy

$ fold pull summarizer

$ vim summarizer.ts

$ fold deploy summarizer.ts --name summarizer

Deployed v2 https://you.fold.run/summarizer

AI code generation is included with Pro and Scale plans. Free tier users can deploy from files and templates.

Start on the free tier. Upgrade to Pro for AI code generation.