Command Palette

Search for a command to run...

UnylyUnyly
Unyly Deploy · beta

From a GitHub repo —
to a live project.

A site or an MCP server at slug.unyly.org in minutes. Push to the branch — it redeploys. No servers, no configs, no DevOps.

01

Connect a repo

owner/name, branch, env secrets

02

We build it

docker, stack autodetect, live logs

03

It goes live

slug.unyly.org + deploy on push

The console lives at deploy.unyly.org

Projects, deployments, logs and settings — in a clean standalone dashboard.

Open the console

Sites

Static, SPA after npm build, Next.js with a runtime. SPA fallback included.

MCP servers

stdio → streamable HTTP automatically. The URL is ready for Claude and ChatGPT.

Auto-deploy

Push to the branch — the project rebuilds. Deploy history and logs in your dashboard.

Isolation

A container with no extra privileges, CPU/RAM limits, secrets stay yours.

MCP

Deploy from your AI agent

Unyly Deploy is itself an MCP server. Connect it in Claude, Cursor or ChatGPT and deploy by talking: "deploy this repo". The agent creates the project, watches the build, edits env and rolls back.

deploy_projectget_logsredeployrollbackset_envadd_domain

Shipping a project? Add a “Deploy to Unyly” button to your README →

MCP endpoint

https://deploy.unyly.org/mcp

transport: streamable HTTP (MCP)

auth: token from unyly.org/account/gateway

works in Claude · Cursor · ChatGPT

FAQ

What can I deploy?

Static sites (index.html), frontends with npm run build (dist/build/out, SPA fallback), Next.js apps and MCP servers in Node or Python. Custom Dockerfiles are supported too.

How does auto-deploy work?

When you create a project we attach a push webhook to the repo. Every push to the selected branch rebuilds and restarts the project automatically.

How are MCP servers deployed?

An stdio server is automatically wrapped into streamable HTTP: you get a slug.unyly.org/mcp URL that connects to Claude, ChatGPT, Cursor and any remote-capable client.

Is it safe?

Every project runs in an isolated docker container with strict CPU and memory limits and no access to anything else. Secrets are passed only to your own container.

Can I deploy from an AI agent?

Yes. Unyly Deploy is itself an MCP server: connect https://deploy.unyly.org/mcp in Claude, Cursor or ChatGPT (with a token from unyly.org/account/gateway) and deploy by talking. Tools: deploy_project, get_logs, redeploy, rollback, set_env, add_domain.