Shed

Your agent built it.Put it somewhere you can send people.

Claude Code, Cursor, Codex. You sign in once, then it is one file and two commands to a live URL.

Free tier at $0. Nothing to buy yet, and we'll email you when it opens.

Ask, and it is running.

  1. 01

    You ask

    In the session where the code was written, in your own words. No dashboard, no new tab, no form.

    You

    Put this somewhere I can send people.

  2. 02

    It works the rest out

    Shed reads the project and settles the runtime, the build and the port it listens on. Nothing is configured by hand, and the build runs on Shed rather than on your laptop.

    $ shed deploy . --remote --wait

    detected node 22, build, port 8080

    building on Shed

    live at https://orders-api.shed.codes

  3. 03

    You say who gets in

    It was private from the moment it existed. Naming someone is what makes the URL work for them, and it never works for anyone else.

    Named people only

    • ann@team.dev
    • tomas@team.dev
    • everyone else

Every host can run it. Count what it costs to get there.

Measured in handoffs, not minutes: the moments the agent stops and waits for you to open something. They are what actually make setup feel long.

the agent can do it from the shell you have to go somewhere else

  1. Google Cloud RunBilling and three APIs before the first deploy.cloud.google.com/run/docs/quickstarts/deploy-container
    1. create a Google account
    2. create a Cloud project
    3. attach a billing account
    4. install the gcloud CLI
    5. gcloud auth login opens a browser
    6. set the active project
    7. enable the Cloud Run API
    8. enable the Cloud Build API
    9. enable Artifact Registry
    10. answer the region and access prompts
    11. deploy

    9handoffs

  2. VercelBuilt around a Git connection and a framework preset.vercel.com/docs/getting-started-with-vercel
    1. create an account
    2. install the CLI
    3. vercel login
    4. link or create the project
    5. confirm the framework preset
    6. deploy

    4handoffs

  3. RailwayThe deploy succeeds and nothing is reachable yet.docs.railway.com/quick-start
    1. create an account
    2. install the CLI
    3. railway login
    4. answer the railway init prompts
    5. railway up
    6. generate a domain in the dashboard

    4handoffs

  4. Fly.ioOne launch command, but it interviews you.fly.io/docs/launch
    1. create an account
    2. install flyctl
    3. fly auth login
    4. add a payment method
    5. answer the launch prompts
    6. deploy

    4handoffs

  5. Cloudflare WorkersNo billing form, but the scaffold interviews you.developers.cloudflare.com/workers/get-started/guide
    1. create a Cloudflare account
    2. install the Wrangler CLI
    3. answer the create-cloudflare prompts
    4. wrangler login opens a browser
    5. deploy, claiming a workers.dev subdomain

    3handoffs

  6. ShedSign in once, then it is one file and two commands.the CLI in this repository
    1. create an account
    2. install the CLI
    3. shed login, approve in the browser
    4. shed init writes the SHED file
    5. shed deploy returns a URL

    2handoffs

Before you hand it off.

What can I actually run on it?+

Web applications and APIs, workers, one-off scripts, and containers with the build and run settings declared. Not a frontend host. Shed builds from what the SHED file declares and nothing else: no ambient laptop state, no hidden dependencies.

Are deployments isolated?+

Yes. Each deployment runs in a gVisor sandbox on a node pool reserved for tenant workloads, so the kernel your code calls into is not the host's. Nothing shares a process tree with anything else you're running.

How does scaling work?+

A Shed scales to zero when idle and back up when a request arrives. The bounds are set by us for now; per-project limits are not exposed yet.

Where does my code run, and how do I see what it is doing?+

Managed Kubernetes on GCP and AWS, each deployment in its own sandbox. Low-cost data centres are planned next, so idle software costs less to keep alive. shed logs <id> streams stdout and stderr from any run.

Why not Railway or Fly.io?+

Their unit is a project you tend: name it, add services, expose it in a second step. Agents produce software nobody tends, in volume. Different unit, not a worse cloud. Railway is a canvas.

Why not Vercel or Cloudflare?+

Vercel starts from a repository and a framework it recognises; an agent has a folder and twenty minutes. Cloudflare has every piece, and expects you to assemble them. Vercel is a workflow, Cloudflare is a kit of parts.

Give your agents somewhere to build.