Questions, answered.

Any MCP client that supports Streamable HTTP. That includes Claude Code, Claude Desktop, Cursor, Codex, VS Code with Copilot, Windsurf, Gemini CLI and opencode. The Connect page in the dashboard has the setup for each.

Both let you limit spending too: Railway has hard usage limits, and Vercel can pause a project when spending reaches an amount you set. On Dawnrose the limit is always on and paid in advance. Past what your plan includes, usage comes out of credit you bought, your apps stop at your limit, and we never charge your card for overage. Only the workspace owner, signed in to the dashboard, can change the limit, never an agent or a token. Your agent works through 76 MCP tools. The ones that delete a project, service, volume or backup are marked destructive, so Claude Code, Cursor and others ask you first, and the server refuses them without an explicit confirm. The activity log names the agent on every change it makes. Railway’s trial credit expires after 30 days; Dawnrose’s $5 doesn’t expire, it lasts until it’s used. Railway and Vercel are established, and Dawnrose is new, with no SLA yet.

No. Everything your agent can do, you can do in the dashboard or with the dawn CLI, which deploys the folder you are in. Install it with npm install -g https://dawnrose.ai/sdk/dawnrose-cli-0.1.0.tgz.

An agent token reaches one workspace, and can do what its owner’s role allows, except manage people: invites, roles and project access are changed only in the dashboard, signed in. Sealed secrets can’t be read back. Destructive tools are marked as such, so the agent’s client asks you first, and the server refuses them without confirm: true. Every action shows up in the activity log with the agent’s name.

No. The trial stops at its $5 of credit. On a paid plan your apps stop when usage passes what the plan includes and you have no credit left, or at your spend limit: $20 on Hobby and $100 on Pro until you change it. We don’t charge your card for extra usage.

CPU and memory by the second, volumes and backups by the hour, and outbound data by the GB. Outbound data is everything leaving your containers, direct connections and WebSockets included. Traffic between services on the same server isn’t billed. A stopped service uses no CPU or memory, but its volume still counts. Builds count too: a build’s CPU and memory are metered the same way while it runs.

Yes. Past what the plan includes, with no credit left, the workspace is suspended within a few minutes and its services stop. The same happens at your spend limit, if that comes first. You get warnings at 50%, 75% and 90% of the period’s budget before that. Add credit or raise the limit and it comes back within a few minutes. On a paid plan it also comes back when the next billing period begins. Services start again on their next request. If others took the memory a sleeping service held on its server while you were suspended, that one is stopped instead and says so: deploy it again and it’s placed where there’s room. A service with a volume stays on its server.

It stays. Containers are stopped, not removed, and volumes, databases and backups are kept. Nothing deletes them automatically. Visitors to your domains get an error, and deploys are refused. Stored volumes and backups keep counting as usage at $0.15 per GB a month.

Your apps stop and deploys are refused. Your data stays. To carry on, buy credit or pick a plan under Billing in the dashboard. Online payments aren’t open yet, so for now email us and we can add credit to your workspace.

No, if you sign up with Google or GitHub: the trial includes $5 of usage, once, with no card. GitHub accounts younger than 30 days don’t get the trial credit. It’s also limited each day, so on a very busy day a new account may start without it. Accounts made with an email and password don’t get the trial credit, and need credit or a plan before they can deploy. While online payments are closed, sign up with Google or GitHub: an account made with an email and password can’t deploy yet.

Memory is set aside on a server whether it’s used or not. A workspace can hold what its balance would pay for over 7 days at $10 per GB a month, or a small app’s worth (2 GB on Trial, 4 GB on Hobby, 8 GB on Pro), whichever is more. Redeploying what already runs, or rolling back, works when money is low.

No SLA yet. We don’t offer one, and we don’t publish an uptime figure or a status page yet.

A few things people ask for. There are no GPUs, so it isn’t the place for model training or GPU inference. You sign in with Google, GitHub or an email and password: there’s no SAML or single sign-on for your company’s directory. There’s no SLA yet. And closing an account goes through support@dawnrose.ai, not a button.

In EU Central (Falkenstein, Germany), EU North (Helsinki, Finland) and US East (Virginia, US). You pick the region when you create a project, and its services run on servers there. Usage costs the same in EU Central and EU North. In US East it costs 1.75 times as much, because servers there cost more. Each app runs in a gVisor sandbox, on a private network shared only with the services in its environment.

Yes. Add it to a service and create the DNS records shown, and the certificate is issued automatically. If the domain’s DNS is on Cloudflare, connect it once and Dawnrose adds the records for you.

n8n, WordPress, Ghost, Metabase, Open WebUI and the rest of the 24 templates, each with its database and an admin account that already signs in. Anything else deploys from a repository or a Docker image.

Yes. The dawn CLI deploys the folder you’re in and can bridge MCP for clients that only speak stdio. Install it with npm install -g https://dawnrose.ai/sdk/dawnrose-cli-0.1.0.tgz (Node.js 22 or later). It’s served from this site, not the npm registry. The CLI docs list every command.

A replica may keep 2 GB on Trial, 10 GB on Hobby and 40 GB on Pro outside a volume. Past that it’s stopped and you get a notification. Its files outside a volume are gone, and you can deploy again when you’re ready, so keep anything you need in a volume. If a whole server passes 90% full, cleanup runs first, then the biggest user’s biggest service is stopped the same way, counting a trial’s use double. Volumes are never touched.

Each service keeps its newest 100 deployments, and a workspace has room for 32 MB of deployment records, variables and activity on Trial, 256 MB on Hobby and 1 GB on Pro. Past either, the oldest finished deployments are removed, which can take old rollback targets with them. Build and deploy logs are kept for 7 days on Trial, 14 on Hobby and 30 on Pro, up to 0.5 GB, 5 GB and 20 GB in total. Images beyond what’s running may take 6 GB, 30 GB and 100 GB on each server, and old rollback targets give way first. The activity log keeps 14 days on Trial, 30 on Hobby and 90 on Pro, up to 4 MB, 16 MB and 64 MB. A running deployment always stays.

Almost anything. Port 25 is closed on every plan, and mail submission ports are open on Pro only. Trial apps can’t connect out to SSH, RDP or SMB ports (22, 23, 135, 139, 445, 3389 and 5900). On Hobby and Pro those ports take 2 new connections a second per container. Speed and connection limits are under each plan.

An app can’t reach the Dawnrose API or dashboard from inside its container. Call them from outside Dawnrose, such as your own servers or CI. To start a deploy from CI, you can also use the service’s deploy hook.

There isn’t a button for it yet. Delete your projects in the dashboard, then write to support@dawnrose.ai from the address you signed up with, and we’ll delete the account and your personal information, as the privacy policy says.

Signed in, use Support at the bottom of the dashboard’s sidebar: the form sends your message with your workspace attached. Or email support@dawnrose.ai. Either way, include the project name and what you expected to happen.