Give your AI coding agent real infrastructure to work with — GitHub repos, Postgres databases, preview environments, custom domains, and a secrets vault. Tell your agent what to build; Deploy Mill handles the rest.
Deploy Mill gives your AI agent a full deployment toolkit — from spinning up repos to managing databases and preview environments. Just describe what you want; the agent handles the rest.
Say "deploy a Node API with Postgres and a custom domain." Deploy Mill creates the GitHub repo, scaffolds the template, runs CI, deploys the container, and wires TLS — all in one shot via start_project.
Each branch gets its own live HTTPS environment with a forked database and fresh storage — so you can test against real data without ever touching production. Previews self-destruct when their TTL expires.
Each app gets its own Postgres database — provisioned automatically when you deploy and wired up as DATABASE_URL without any manual steps. Deleted cleanly when you delete the app.
Sensitive values — API keys, tokens, credentials — are entered by you in a secure browser link and stored encrypted. Your AI agent can inject them into apps by name but can never read them back.
Your agent attaches a domain, Deploy Mill provisions a free TLS certificate and tells you exactly which DNS record to add if it's not pointed yet. No dashboard clicking required.
When a deploy fails, your agent can fetch the build and runtime logs directly — read the error, fix the code, and redeploy, all without you having to look at a dashboard.
Every deploy is tagged. If something goes wrong, your agent can swap back to the previous working image in seconds — no rebuild, no downtime, no data risk.
Declare named storage volumes for caches, indexes, or uploaded files. They survive every redeploy automatically — no manual backup or restore needed.
Deploy web apps (Node, Python, static) or headless background workers — queue consumers, scheduled jobs, data pipelines — using the same workflow. No separate config needed.
Scaffold a new project from a template, adopt an existing GitHub repo, or push files directly via the API. Your agent can also get short-lived, scoped git credentials to clone and push.
Stop an app's container without deleting anything — your repo, env vars, domains, and database all stay intact. Start it back up in seconds when you need it again.
Every team gets its own isolated workspace. Your apps, secrets, and databases are never shared with other users. Sign in with Google or email — no static API keys to manage.
Traditional deploy tools were built for humans clicking dashboards. AI agents need structured outputs, clear errors, and the ability to recover from partial failures — that's what Deploy Mill was built for.
| Capability | Deploy Mill MCP | Raw APIs / shell scripts | PaaS dashboards |
|---|---|---|---|
| Typed structured tool outputs | ✓ Discriminated unions | — Raw JSON / exit codes | — Human UI only |
| Partial-failure recovery | ✓ failedAt + resume |
— Manual retry logic | — Start over |
| Isolated preview envs | ✓ Forked DB + volumes | ~ DIY infra | ~ Some PaaS |
| Secrets never visible to agent | ✓ Human-entry vault | — Env vars in prompts | ~ Varies |
| Agent-driven DNS + TLS | ✓ Fully automated | — Human clicks required | — Dashboard only |
| Image-swap rollback in seconds | ✓ Single tool call | ~ Requires setup | ~ Varies |
| Provider-neutral config schema | ✓ Swap backends later | — Tied to provider | — Tied to provider |
Deploy Mill routes through battle-tested providers. Swap any of them later — tool signatures stay the same.
Deploy Mill was designed from the ground up so AI agents can never escalate beyond their workspace, and secrets never flow through the tool surface.
Sensitive values enter via a single-use human browser link and are stored encrypted. bind_secret injects by reference — the API surface never returns a secret value. Human-only entry, always.
Every workspace is a separate tenant. Apps, secrets, databases, and domains are scoped to the org. OAuth tokens can never cross workspace boundaries. No shared infrastructure between tenants.
Authentication uses Better Auth with Dynamic Client Registration and PKCE. Each MCP client gets a short-lived token. The Dokploy API key is internal — MCP clients never see it.
get_clone_credentials mints short-lived, repo-scoped tokens for direct git access. Credentials expire automatically. Your GitHub App token never leaves the server.
No CLI to install, no YAML to write. Sign up, connect your agent, and start deploying.
Sign up with email or Google in seconds. You get an isolated workspace — your own apps, databases, secrets, and domains, never shared with anyone else.
Add Deploy Mill's MCP endpoint to Claude Code, Codex, or any compatible AI agent. The OAuth handshake happens automatically — no API keys to copy and paste.
Just describe what you want. Your agent handles the GitHub repo, template, CI pipeline, container, database, and domain automatically. If anything fails midway, it knows exactly where to resume.
Spin up a full preview environment for every branch — its own database, storage, and HTTPS URL. Test against real data without ever touching production. When you're done, one command tears it all down cleanly.
Free to start, no credit card required. Connect Claude or Codex and your AI agent can deploy real apps, databases, and preview environments — without you babysitting it.
Create your account