Skip to content

fix(@wdio/session): keep agents on the UI path and fail fast - #1

Closed
Winify wants to merge 1 commit into
vg/snapshot-row-contextfrom
vg/session-bench-fixes
Closed

Winify wants to merge 1 commit into
vg/snapshot-row-contextfrom
vg/session-bench-fixes

Conversation

@Winify

@Winify Winify commented Oct 7, 2026

Copy link
Copy Markdown
Owner

Changes driven by a review of failed Online-Mind2Web benchmark runs:

  • click: a hidden element's error no longer prints its href, so agents reveal it instead of navigating to the URL and skipping the UI
  • click: a covered-by error names the dismiss control of the covering dialog or cookie banner, e.g. Close it first: button "Accept all".
  • launch Chrome and Edge with --disable-blink-features=AutomationControlled so navigator.webdriver is false (not sent to cloud sessions)
  • read and find take --offset to page through long output
  • read --scope scrolls the element into view, so the screen shows what the agent read
  • a timed-out action suggests checking the page and reloading instead of a longer --timeout, and --timeout is capped at 60 s except for wait
  • exec hints the v10 getHTML options when a boolean is passed; the skill says exec runs in Node

Proposed changes

How you tested

  • Unit tests for the touched package(s)
  • Type definition tests (if the public API / command types changed)
  • Smoke or e2e (only if the testrunner, launcher, or a real browser path changed)

Types of changes

  • Polish (an improvement for an existing feature)
  • Bugfix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Documentation update (improvements to the project's docs)
  • Specification changes (updates to WebDriver command specifications)
  • Internal updates (everything related to internal scripts, governance documentation and CI files)

Checklist

  • I have read the CONTRIBUTING doc
  • I have added tests that prove my fix is effective or that my feature works
  • I have added the necessary documentation (if appropriate)
  • I have added proper type definitions for new commands (if appropriate)

Backport Request

  • This change is solely for v10 and doesn't need to be back-ported
  • Back-ported PR at #XXXXX

Further comments

Reviewers: @webdriverio/project-committers

Changes driven by a review of failed Online-Mind2Web benchmark runs:

- click: a hidden element's error no longer prints its href, so agents
  reveal it instead of navigating to the URL and skipping the UI
- click: a covered-by error names the dismiss control of the covering
  dialog or cookie banner, e.g. `Close it first: button "Accept all".`
- launch Chrome and Edge with --disable-blink-features=AutomationControlled
  so navigator.webdriver is false (not sent to cloud sessions)
- read and find take --offset to page through long output
- read --scope scrolls the element into view, so the screen shows what
  the agent read
- a timed-out action suggests checking the page and reloading instead of
  a longer --timeout, and --timeout is capped at 60 s except for wait
- exec hints the v10 getHTML options when a boolean is passed; the skill
  says exec runs in Node
@Winify Winify closed this Oct 8, 2026
@Winify
Winify deleted the vg/session-bench-fixes branch October 8, 2026 19:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant