A request. A running plan.
How does an agent host an application on Euphoric?
It talks to one HTTP API from start to finish: it registers a credential it generated itself, adds prepaid credit, sets a spending ceiling and submits a container image. Euphoric creates a virtual machine for that application and reports each step as an operation the agent polls until it finishes. The journey below is implemented and was verified in private tests on real servers in the EU; it is not open to the public yet.
Bring a credential and a contact
Generate a bearer credential and register it with an email address or HTTPS callback. There is no signup form or password. Contact starts unverified; verifying it is a separate, explicit step.
Fund a bounded runtime
Add prepaid credit, then give each application a lifetime spending ceiling and, optionally, a stop time. Runtime is reserved ahead in small windows, so an application cannot run past what is funded.
Deploy and follow the operation
Submit a digest-pinned container image, such as a web service or a remote MCP server. The API answers with an operation ID straight away; poll it until it succeeds, fails or is cancelled. After a timeout, repeat the same request with the same idempotency key and get the same answer.
Inspect, suspend, export, delete
Read status, operation events and container logs. When credit runs out, the application is suspended and kept for 72 hours. Export returns its configuration. Deletion releases its reservations only after cleanup is confirmed.
A person can take over later
Planned: a human claims the application, keeps its address and data, and gets a normal invoice; the agent's credential stops working. Customer accounts, ownership claim, source builds, databases and custom domains are not built yet.
Follow the complete API walkthrough or read the planned pricing.