For customersBeta

Computer use with CUA

Auction-routed inference with an explicitly paired local desktop driver across the SDK, API, and web app.

Owner
Product
Version
beta-2026.07
Verified
2026-07-13

Omnious computer use combines auction-routed inference with a local, explicitly paired CUA Driver. The router chooses and serves the model; the local adapter alone holds Accessibility, Screen Recording, mouse, and keyboard capability.

the boundaryThe hosted Omnious API never starts CUA Driver and cannot operate your desktop. It receives tool schemas, returns model tool calls, and routes later screenshots or accessibility results only when your local client sends them back for inference.

One capability, two surfaces

SurfaceWhat runs locallyWhat the API does
SDKCuaDriverAdapter owned by the callerReturns tool calls; never executes them
APIYour application's executorValidates intent and records the tool policy

The curated tool surface

CUA Driver has a broad low-level surface. Omnious exposes five stable facade tools so provider context stays small and maintenance capabilities cannot become model-callable accidentally.

ToolIntentBehavior
computer_healthreadDriver, platform, and permission diagnostics
computer_listreadApplications or windows for one process
computer_snapshotreadFresh accessibility tree plus grounding screenshot
computer_actionexecuteOne approved click, type, key, scroll, drag, value, cursor, or app-launch action
computer_sessionexecuteNamed cursor and cleanup boundary for one run

Every window-targeted action is bracketed by a fresh pre-action snapshot and a post-action verification snapshot. Element indices are never reused across snapshots. App launch has no pre-existing window to snapshot, so the explicit approval is its verification boundary. Foreground delivery remains an escalation, not a default.

SDK

caller-owned adapteromnious-sdk
import { CuaDriverAdapter, OmniousClient } from "omnious-sdk";

const computer = new CuaDriverAdapter({
  approve: async ({ action, arguments: args }) => {
    // Replace with your own product approval UI.
    return await showApproval({ action, args });
  },
});

const omnious = new OmniousClient({
  routerUrl: "https://api.omnious.xyz",
  creditKey: process.env.OMNIOUS_CREDIT_KEY,
});

try {
  const result = await omnious.ask("Inspect the selected app", {
    model: "auto",
    intent: "execute",
    computer,
  });
  console.log(result.content);
} finally {
  await computer.close();
}

The approval callback is mandatory for actions: omitting it fails closed. Multimodal CUA results remain image blocks throughout Chat, Responses/Codex, and Anthropic Messages tool loops.

Permission policy

Omnious regenerates a fixed, private CUA policy before every driver process. It permits only the native calls needed by the five facade tools. Browser JavaScript execution, app killing, forced foregrounding, recording, replay, configuration changes, and dependency installers remain denied.

privacy and costA screenshot or accessibility tree used for reasoning passes through Omnious and the winning inference provider, and contributes to input usage. Pair only a machine and app session whose visible content you are willing to send. Passwords, seed phrases, API credentials, private keys, and unrelated signed-in applications should stay out of scope.
API callersRaw API clients may send these same function schemas, but execution remains their responsibility. routing.intent: "read" or"plan" rejects computer_action before auction or payment with tool_intent_conflict.