Expand control.mdx with computer use + playwright execution patterns

[masnwilliams]

Summary

Stacks on top of #370. Adds three named patterns as short bullets under the existing "Computer use + playwright execution" section in introduction/control.mdx:

• Expose it as a tool — model calls playwright.execute for structured data, the agent stays in charge of navigation.
• Checkpoint between steps — assert cookies / DOM state after a CUA step before continuing.
• Drop the agent once selectors are stable — replay at scale without a model in the loop.

No new page, no extra code. Just names + one-sentence explanations so customers asking "how do I combine these?" have something concrete to point at.

Test plan

• mintlify dev renders control.mdx with the new bullets
• No broken links

---

Note

Low Risk
Low risk documentation-only change that adds guidance text without modifying any runtime code or APIs.

Overview
Expands introduction/control.mdx by adding a short "A few ways this shows up in practice" subsection under Computer use + playwright execution.

The new bullets describe three recommended integration patterns: using playwright.execute behind an agent tool, checkpointing with DOM/cookie assertions between computer-use steps, and switching to Playwright-only automation once selectors are stable.

^{Reviewed by Cursor Bugbot for commit 8ba5164. Bugbot is set up for automated code reviews on this repo. Configure here.}

[firetiger-agent]

Firetiger deploy monitoring skipped

This PR didn't match the auto-monitor filter configured on your GitHub connection:

Any PR that changes the kernel API. Monitor changes to API endpoints (packages/api/cmd/api/) and Temporal workflows (packages/api/lib/temporal) in the kernel repo

Reason: PR only adds documentation (patterns.mdx guide) with no changes to API endpoints or Temporal workflows in packages/api/.

To monitor this PR anyway, reply with @firetiger monitor this.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit e9444a4. Configure here.}

[cursor]

Stale comment

Risk assessment: Very Low

I reviewed the synchronized diff directly. The PR adds six lines of documentation guidance to introduction/control.mdx under an existing section. There are no runtime code changes, docs config changes, infrastructure changes, auth/permission changes, schema changes, or shared production logic changes. git diff --check passed for the requested diff.

No CODEOWNERS file is present in the repo. The PR already has an approval from the earlier assessment, and this update does not increase risk, so I’m leaving that approval in place without re-approving.

  

_{Sent by Cursor Automation: Assign PR reviewers}

[cursor]

Risk assessment: Very Low

I assessed the actual diff for cf6c7bbe2762ba139ecfff086319f2fe061e0c7b...8ba5164cc60cde0f2e3f87e8dc7e0582bc4f49cd. The PR adds six lines of documentation guidance to introduction/control.mdx and does not modify runtime code, docs configuration, infrastructure, auth/permissions, schemas, or shared production logic.

Validation: git diff --check passed for the PR diff. I found no CODEOWNERS file in the repo. GitHub already reports the PR as approved, and this update does not increase risk, so I’m leaving the existing approval in place without submitting a duplicate approval.

  

_{Sent by Cursor Automation: Assign PR reviewers}