Can My Agent Uselookup · compare · cite

Aug 28, 2026 · Catalog foundation is up

Atomic capability · runtime

Computer useProduct capability

Inspect and act on a visual interface through screenshots and input actions.

Computer use: 1 supported, 4 partial, 0 unsupported, 26 unreviewed across 31 cataloged products.

Markdown · JSON

Current evidence by product

Can my agent use Computer use?

Read across for the answer. 5 of 31 current product columns have reviewed evidence; unreviewed does not mean unsupported.

Known issues (0)Report an issue
  • Supported1
  • Partial4
  • Unsupported0
  • Unknown26
  • Not applicable0

Web

9 products

Desktop

13 products
CursorCursor
P2Partial
Docs checked 2026-08-28Current · 2 conditions
?Unknown
No source reviewedPreview record
Report this result

CLI

9 products

Unknown means no public evidence has been reviewed for that product and capability. It does not mean unsupported.

How statuses are assigned

Definition and scope

What this capability means

This row asks whether the exact harness can inspect a rendered visual state and perform direct input actions. It distinguishes broad computer or GUI control from a browser-only feedback loop.

That distinction matters: a coding harness can fully automate its integrated browser while leaving desktop applications and operating-system controls outside scope. Plan, region, policy, and runtime boundaries remain attached to each sourced cell.

Traceable compatibility

Assertion ledger

Documentation evidence only. No runtime conformance test is implied.

ChatGPTweb · current
Partial
Target
2026-08-28 ChatGPT Work documentation observation · hosted-observation
Environment
hosted-default
Observed
2026-08-28
  • planpaid ChatGPT plan required
  • regionsupported regions only
  • policyworkspace permission and site restrictions apply
Evidence
Cursordesktop · current
Partial
Target
current Cursor browser documentation · dated-documentation
Environment
local-default
Observed
2026-08-28
  • runtimeintegrated browser only
  • policyenterprise MCP origin policy can restrict access
Evidence
Copilot Chatdesktop · current
Partial
Target
current VS Code browser-tools documentation · dated-documentation
Environment
local-default
Observed
2026-08-28
  • runtimeintegrated browser only
  • policyorganization policy can disable browser tools
Evidence
OpenWork Desktopdesktop · current
Partial
Target
current OpenWork Browser documentation · dated-documentation
Environment
local-default
Observed
2026-08-28
  • runtimefirst-party OpenWork Browser only; no general desktop-app control
  • policyOpenWork Browser extension must be enabled
Evidence
Grok Botdesktop · current
Supported
Target
2026-08-28 Grok Bot desktop documentation observation · hosted-observation
Environment
hosted-default
Observed
2026-08-28
  • runtimecomputer use runs on an account-scoped hosted Linux VM; all Bots for the account share its files, sessions, and credentials
  • policyone Bot can run one computer-use task on its screen at a time; sensitive steps should use operator takeover
Evidence
  1. 1. Evidence checked 2026-08-28: ChatGPT Work can use a separate cloud browser to read pages, click, enter form data, and continue supported tasks, but access varies by paid plan, region, rollout, and workspace permission.
  2. 2. Evidence checked 2026-08-28: Cursor Agent can navigate, click, type, scroll, and capture screenshots in its integrated browser; this is browser control, not general operating-system GUI control.
  3. 3. Evidence checked 2026-08-28: VS Code agents can navigate, read, screenshot, click, type, hover, drag, and handle dialogs in the integrated browser; organization policy can disable the tools.
  4. 4. Evidence checked 2026-08-28: OpenWork Desktop's first-party browser can open pages, click, fill forms, read page content, and take screenshots; OpenWork explicitly distinguishes this from general operating-system control.
  5. 5. Evidence checked 2026-08-28: Grok Bot exposes a persistent hosted computer with browser and desktop tools. The operator can watch clicks, typing, and navigation, take over sensitive steps, and let work continue after closing the preview.