Reference architectures

What the platform can compose today.

No invented logos or vanity numbers. These reference architectures show how current Korve runtime and data primitives fit together for concrete application workloads.

Six practical application systems

Each example is a product composition, not a claim that a named customer runs it in production.

Agent-built internal tool

Deploy the UI and API, protect the application, and let a scoped project operator investigate failures with approval gates.

Review application

Create an ephemeral application environment for a pull request and remove it at the end of the review lifecycle.

Transactional SaaS

Combine an always-on application, managed Postgres, object storage, recovery policy, and database advisor evidence.

Webhook ingestion

Receive requests in the application, publish asynchronous work to a queue, and route failed attempts to a dead-letter workflow.

Scheduled reporting

Use a cron target to enqueue work, store generated output, and inspect application logs and runtime metrics.

Controlled customer portal

Pair application auth with an explicit access policy, custom domain, resource-scoped secrets, and project audit history.

Workflow

A composition you can inspect end to end

  1. 01

    Choose the system boundary

    Identify the app runtime and the minimum data and operations primitives required by the workload.

    Resources named
  2. 02

    Automate through one spec

    Use the same resource operations from the dashboard, API, Rust CLI, MCP, or infrastructure manifest.

    Contract shared
  3. 03

    Operate the whole project

    Use deployment history, logs, metrics, budgets, recovery state, and operator runs to understand changes.

    Evidence retained

Give your agent a cloud it can operate.

No free tier. No seats. Pay for what you run.