Automations don't support Computer/Browser Use

Ideas Feature Requests
, , ,
1

Where does the bug appear (feature/product)?

Background Agent (GitHub, Slack, Web, Linear)

Describe the Bug

New Automations Bug:

I’ve created a Automation where I explicitly ask the agent to take a screenshot of the before and after but the Agent in the Automation context, doesn’t do it. It doesn’t seem to have access to computer use like a normal agent.

This is ruining it for the purposes of taking a bug from linear, re-creating it and then solving, screenshooting the solution. Which I assume would be a common use case.

I took the exact same prompt and put it into the normal agent and it produced screenshots fine, so seems to be Automation specific.

Steps to Reproduce

  1. Use Prompt Example in Automation (working in Agent, not in Automation):

You are a Senior Developer for AutoRFP.ai triaging bugs from Linear.

Workflow:

  1. Read the bug from Linear. Understand it fully before touching code.

  2. Research the root cause in code. Trace the component tree, read the relevant files, identify the root cause mechanically. Do not start services or open the UI yet.

  3. Implement the fix. Minimal diff only — follow existing patterns. Do not refactor or add unrelated changes.

  4. Start services and set up test data. Use bash .cursor/scripts/start.sh then start Router and Frontend per AGENTS.md. If the bug requires special data, script it — don’t set it up through the UI.

  5. Capture before and after. Stash or revert your fix, take a quick screenshot of the bug, re-apply the fix, take a screenshot of it resolved.

  6. Keep this to one pass — if the environment fails to load on the first attempt, fall back to code-level before/after and note the issue.

  7. Commit, push, document. Single commit with a conventional commit message. Add a concise Linear comment covering: root cause, fix applied, before/after, and branch name.

Rules:

  • Do not start services until the fix is written. Code-first investigation is faster.
  • Pick the best solution and implement it. Only mention alternatives if there are genuine trade-offs worth flagging.
  • If the bug is obvious from code reading, skip deep investigation — fix it, verify, done (with evidence).
  • If unsure about the fix or its impact, flag concerns on the Linear issue rather than spending time speculating.

The linear issue should be kept in the Triage status, change it back to that status once the PR is up.

  1. Submit Bug against workflow.

  2. Observe the completed run that there is no Computer Use or Demonstration/Screenshots

Expected Behavior

It should produce a screenshot/demonstration as specifically asked.

Operating System

MacOS

Version Information

This was done through the WEB UI, so below is N/A:

Version: 2.6.13
VSCode Version: 1.105.1
Commit: 60faf7b51077ed1df1db718157bbfed740d2e160
Date: 2026-03-06T06:17:49.499Z
Build Type: Stable
Release Track: Nightly
Electron: 39.6.0
Chromium: 142.0.7444.265
Node.js: 22.22.0
V8: 14.2.231.22-electron.0
OS: Darwin arm64 25.2.0

For AI issues: which model did you use?

Opus 4.6 High

Additional Information

When asking the Automation started Agent if it has access to a computer/browser it says no.

Does this stop you from using Cursor

No - Cursor works, but with this issue

1 Like
3

Hey, thanks for the detailed report. Right now, Computer Use (screenshots, browser interaction, virtual desktop) isn’t available in the Automations context. The tools available for Automations are: Open PR, Comment on PR, Request Reviewers, Send to Slack, Read Slack channels, MCP servers, and Memories. Full list here: Automations | Cursor Docs

That said, your use case (triage a Linear bug, reproduce it, take before and after screenshots, fix, push) is a really strong reason to add this.

As a workaround for now, you can manually start a regular Cloud Agent (from web, desktop, or Slack) with the same prompt. Those do support full Computer Use.

Let me know if you have any questions.

1 Like
4

+1 to get this feature added. We’d ideally like to setup automations to monitor a slack channel, pick up any tickets posted, and attempt to complete them. A big use case for the cloud agents is having it be able to test the changes in the browser, take screenshots/recordings, and “prove” to us (and itself) that it works. Would be nice to have it open the PR, including some of those screenshots as well.

5

Thanks for this, and yeah, might even have Cursor message someone “Please put the following prompt into Cursor” every time a big is in triage, haha.

But yeah, I feel like computer use is even better for automation than the normal agent, as it’s the one time I actually won’t be at a computer to use it and that’s kinda the point.

Appreciate the help in the meantime and would love to see it in the Automations soon1

1 Like
6

This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.