Skip to content
chalupa
ProductAgentsDocsPricing
Sign inInstall the CLI
Menu
ProductAgentsDocsPricingSign inInstall the CLI
chalupa

A disposable copy of the Compose services you pick.

DocumentationAgentsPricingGet in touch

Bring your own cloud. Leave with your data.

Provider costs are estimates. Storage has its own lifecycle.

Documentation

From setup to your next test run.

Run a disposable copy of the Compose services you pick, give an agent an environment, or open a model on a GPU. Find the steps and help for what you are doing now.

/

33 guides•Your own cloud account•SSH is the only open port

Start with your task

What would you like to do?

Follow these four in order. Each one ends with the next step.

01

Run a test on a disposable environment

Install, preview for free, launch, tunnel, test, tear down.

Open guide
02

Copy a recipe for your stack

Starter chalupa.yml files for web and database stacks, queues and search, and replica sets.

Open guide
03

Give your agent an environment

Connect an MCP client, read the action contract, and keep billable actions behind a confirmation.

Open guide

Run and test

Configure the environment, run suites with chalupa test, keep data between runs, and handle secrets.

chalupa.yml referenceEvery field of the config, with a table and an example for each.LifecyclePrepare, launch, work, and sink.Run a suite with chalupa testLaunch, run a declared browser-flow suite from your machine, collect results and sink, in one command.Declarative CIDeclare test commands and run them in a disposable environment.Data and seedsSeparate persistence and repeatable restores.
04

Find your logs and evidence

See what each run leaves behind, what works today and what each Beta piece still lacks.

Open guide
Tunnels and securityPrivate services, local access, supervised tunnels, and secrets.
Expiry and the local teardown runnerHow a session deadline becomes a destroyed server, and how to extend or renew a session.
Secrets with tvaultLeast-privilege injection without files or values in logs.
Local CLI and operator UIInspect and operate a Chalupa from an offline-first terminal.
CLI command referenceThe chalupa command table: each command's flags and behavior.
Copies and variantsRun several copies of one chalupa.yml side by side, each with its own box, ports and volumes.
URLs, routes and sharesWhere a forwarded service answers on this machine, its browser route through the local proxy, and an opt-in remote share.
Logs and diagnoseRead a running box's logs and a bounded diagnosis over SSH, and join a failed test run with the box's own evidence.

Results and evidence

Read results, logs, performance windows and the hosted console.

Suites and artifactsResults, comparisons, and evidence references.Private log artifactsBounded direct uploads, commit verification, and retention.Performance telemetryDurable Monitor windows, privacy boundaries, and charts.Hosted consoleHow heartbeats reach the hosted console, and what each console page shows.

Speed and cost

Make servers ready faster, understand costs, and clean up.

Make servers ready fasterWhere boot time goes, what a prebaked image changes, how to check readiness, and how to reuse a server, a cache and a build across runs.Warm boxes and reuseKeep a server for a short window after the work and reuse it for the next run.Costs and idle observationEstimates and the informational inactivity window.Plans, billing, and managed hoursHosted plans, Stripe lifecycle, credits, and provider-cost boundaries.

Models on a GPU

Run an open model on a GPU and connect your coding agent. Experimental.

Quickstart: managed GPUSet up a managed GPU, run a local coding agent, and stop compute.Remote GPU inferenceExperimental host Ollama through a local SSH tunnel.Agent harnessesConnect OpenCode, oh-my-pi, or Crush to your GPU model.Managed GPUPrepaid GPU balance, rates, and beta availability.

Reference

Concepts, troubleshooting, and exact contracts.

ProvidersSupported providers, billing units, and planned GPU adapters.ConceptsThe mental model: inference, compute, and data.TroubleshootingSafe diagnosis for the most common failures.API and schemaPublic ingestion and read contracts.