The Prompting Company

Benchmark report

Task verified

4m 11s elapsed

The saved attempt met the documented success criterion.

Experience score

93/ 100

Agent-Ready
  • AInstall
  • SAuth
  • SExecute
  • ADocs
  • ACurrent
Experience measures friction in the tested task. The verdict says whether it succeeded.
Agent profile for this run: Install A, Auth S, Execute S, Docs A, Current AAInstallSAuthSExecuteADocsACurrent
Agent profile · this run
Judge's note · what slowed the agent most
The PEP 668 externally-managed-environment error on the first pip install, forcing a venv rebuild before Playwright could be imported.

Where the time went

4:12 agent time
  • Research0:22
  • Attempt2:31
  • Verify1:19
4m 11sTotal elapsed
1:39Time to first success
0:00Waited on the owner

Judge reason

Saved code (browseruse_form_task.py) and live run output prove the full flow. POST https://api.browser-use.com/api/v4/browsers returned HTTP 201 with id e80937ba-e0a1-46e3-8c68-06b8f7f6ed57 and cdpUrl https://e80937ba-....cdp.browser-use.com (evidence/session_create_response.json). Playwright connect_over_cdp passed the key only in the X-Browser-Use-API-Key header (never in a URL or query string); the remote Chrome/154 UA, Windows platform and origin 65.86.109.63 in the httpbin echo confirm a real hosted browser, not local or mocked. page.fill/page.click drove https://httpbin.org/forms/post (no Browser Use agent/task API), and evidence/httpbin_echo.json shows form.custname='Benchmark Test', form.custtel='555-0100', matching the printed JSON byte-for-byte. The finally block issued PATCH /api/v4/browsers/{id} {"action":"stop"} -> HTTP 200 with status 'stopped' and finishedAt 2026-11-03T22:30:27Z (startedAt 22:30:23, consistent). grep for 'bu_' and credential query params across all saved files returned nothing. Transcript idx 56-59 shows the single live run; no fabricated or hard-coded responses.

Attempt record

  1. 0:54Answered from saved key: Browser Use API keyno wait
  2. —First success verified

Attempt artifacts saved.