The Prompting Company

Benchmark report

Task verified

5m 15s elapsed

The saved attempt met the documented success criterion.

Experience score

95/ 100

Agent-Ready
  • SInstall
  • SAuth
  • SExecute
  • ADocs
  • ACurrent
Experience measures friction in the tested task. The verdict says whether it succeeded.
Agent profile for this run: Install S, Auth S, Execute S, Docs A, Current ASInstallSAuthSExecuteADocsACurrent
Agent profile · this run
Judge's note · what slowed the agent most
Doc fragmentation: create-session lacked the endpoint/headers and auth.md was unreachable, so the agent needed three extra fetches plus a search before writing code.

Where the time went

5:15 agent time
  • Research0:22
  • Attempt3:13
  • Verify1:40
5m 15sTotal elapsed
2:21Time to first success
0:00Waited on the owner

Judge reason

Saved code (work/evidence/anchor_form_task.js) calls the documented POST https://api.anchorbrowser.io/v1/sessions with the key only in the anchor-api-key header; session_create_response.json shows HTTP 200 with id b9dc3446-9efb-4c51-b085-415a4cdae995 and cdp_url wss://connect.anchorbrowser.io?sessionId=...&rt=3ab8e650. The script strips any credential query param and connects via playwright-core chromium.connectOverCDP(url, {headers:{'anchor-api-key':...}}); run_output.txt (matching the transcript tool result at the single run) shows '[connect] connected', load of https://httpbin.org/forms/post, and the echoed httpbin JSON with form.custname='Benchmark Test', form.custtel='555-0100', origin 3.142.238.206 and a remote Chrome 136/macOS UA, confirming a real hosted browser rather than a local or mocked one. Termination runs in a finally block via DELETE /v1/sessions/{id}: HTTP 200 OK, body {"data":{"status":"success"}} (session_terminate_response.json). Printed JSON matches httpbin_echo.json byte-for-byte; grep over the saved artifacts finds no credential-like string (only the session UUID and the session-scoped rt token), and the agent's own key grep reported no leak. Nothing mocked or hard-coded.

Attempt record

  1. 1:48Answered from saved key: ANCHOR_API_KEYno wait
  2. —First success verified

Attempt artifacts saved.