> ## Documentation Index
> Fetch the complete documentation index at: https://docs.omnara.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Tenki

> Full Linux computers for your AI agents, running in under a second

[Tenki Sandbox](https://tenki.cloud/products/sandbox) gives your AI agents full Linux computers: hardware-isolated Ubuntu 24.04 VMs, each with its own kernel. Connect your Tenki workspace with a custom pool.

## Why Tenki

* **A real Linux computer.** Each machine runs systemd, and agents get passwordless `sudo`, so they can install Docker or any other package they need.
* **Start with your tools installed.** Tenki's base image comes with Claude Code, Codex, and opencode, plus Python, Node.js, Bun, `git`, `gh`, and build tools. Or define your environment once in a [Tenki template](https://tenki.cloud/docs/sandbox/templates) and launch from it every time.
* **Your custom image, running in under a second.** Each machine's disk mounts straight from Tenki's storage cluster, so nothing is copied before boot.
* **Owned end to end.** Tenki runs its own compute, storage, switches, power, and uplink, so it controls the whole path from request to running machine.
* **Long-running agents.** Tenki keeps each machine running, with no time limit, until Omnara deletes it.

## Set up a Tenki pool

<Steps>
  <Step title="Create an API key">
    In the [Tenki dashboard](https://app.tenki.cloud), open **Tenki Sandbox**, go to **API Keys**, and click **Create API Key**. The key selects the workspace.

    Store the key as an org-owned [secret](/organization/secrets) in Omnara.
  </Step>

  <Step title="Choose an image">
    Leave the image empty to use Tenki's Ubuntu base image with a 20 GB disk.

    For your own tools, build a [Tenki template](https://tenki.cloud/docs/sandbox/templates) with the Tenki CLI and use the image reference it prints, such as `acme/agent-box:v1`. Leave `runtime` out of the template, since Omnara supplies its own start command. Custom images keep the template's disk size unless the pool sets `disk_size_gb`.
  </Step>

  <Step title="Create the pool">
    On the **Machines** page, click **New pool** and choose **Tenki**. Enter the credential, image (or leave it empty), CPU and memory, and capacity limits.

    Only environment variables whose names use letters, digits, and underscores, and don't start with a digit, reach the startup script. The environment and startup script together must stay under about 128 KB.

    To use the API instead, call [Create a pool](/machines/pools#create-a-pool) with the fields under [API fields](#api-fields). The dashboard can't set the disk size, so set `disk_size_gb` through the API.
  </Step>
</Steps>

## API fields

When you create a pool through the API, send these fields in the [Create a pool](/machines/pools#create-a-pool) request. `provider_config` applies to the whole pool, and `default_machine_provider_options` sets the options for each new machine, which project grants and agent configs can override. For example:

```jsonc theme={null}
{
  "name": "tenki-pool",
  "provider": "tenki",
  "provider_auth_secret_id": "sec_hcyd5n6a2bfgik7mv4qtrwz3je",
  "default_machine_provider_options": {
    "disk_size_gb": 40 // optional
  },
  "default_machine_cpu": 2,
  "default_machine_memory_mb": 4096,
  "max_machine_cpu": 4,
  "max_machine_memory_mb": 8192,
  "max_total_cpu": 20,
  "max_total_memory_mb": 40960,
  "max_total_machines": 10
}
```

* `image` is optional. Omit it, as here, to use Tenki's base image, or set a Tenki image reference such as `acme/agent-box:v1`.
* `disk_size_gb` ranges from 5 to 100. It defaults to 20 for the base image and to the template's size for custom images.
* Memory must be a multiple of 2 MB.
* Optional: `startup_script` in `default_machine_provider_options` runs on each new machine before the daemon starts, and `allowed_images` in `provider_config` is an [allowlist](/machines/providers/overview#allowlists).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.