shiply.now

Getting Started

Publish to the web in seconds from any AI agent

shiply.now is instant web hosting built for AI agents — static or server-rendered. Any tool that can make HTTP requests can publish files and get a live URL at <slug>.shiply.now in seconds — no account required to start.

The 60-second tour

  1. Tell your agent to publish. Paste the agent prompt from the homepage (or your dashboard) into Claude Code, Cursor, Codex, or any LLM with a shell. The agent installs the shiply skill on first run and learns the API.
  2. Your site goes live. First publish returns a live URL plus a claimUrl. Open the URL — the site is on Cloudflare's edge already.
  3. Keep it forever. Open the claimUrl (or, if your agent printed a https://shiply.now/auth/XXXX-YYYY link, click Allow on that consent page). The site lands in your dashboard and the agent gets a permanent API key for future publishes — one click, then it's instant from then on.

Where to go next

  • Quick start — the literal first publish, with copyable commands
  • Authentication — API keys, the device flow, agent onboarding
  • Install the CLInpx -y shiply-cli skill
  • MCP — connect Claude / Cursor / Codex to shiply via the MCP server at https://shiply.now/mcp
  • Sites — updates, SPA mode, rollbacks, site management
  • API reference — every endpoint, with examples

What you can publish

Static files (HTML, CSS, JS, images, PDFs, video, JSON, fonts) serve straight from the edge. Need server-side rendering? shiply deploys SSR frameworks as a per-site Cloudflare Worker. For lighter dynamic behaviour on a static site, use Proxy routes (server-side secret injection), Site Data (form/waitlist collections), or Databases (per-site Cloudflare D1 or Neon Postgres).

Every major framework deploys — static or server-rendered: Next.js (via OpenNext), SvelteKit, Astro, Qwik, Nuxt and the Nitro family, React Router v7, Vite, Hugo, Jekyll, Eleventy, and more — see Frameworks for the exact build commands per stack.

How publishing works under the hood

The publish flow is three steps so partial failures can't leave a broken site live:

  1. POST /api/v1/publish with a file manifest → returns presigned upload URLs
  2. PUT each file's bytes to its presigned URL
  3. POST the finalize URL → site flips live atomically

Both the CLI and the MCP server handle this for you. Raw API is in the reference.