The Prompting Company

Benchmark report

Task verified

4m 6s 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
The official domains docs page fetched as a bare endpoint index with no fields, and the async DELETE forced an extra polling script to confirm removal.

Where the time went

4:07 agent time
  • Research0:29
  • Attempt2:13
  • Verify1:24
4m 6sTotal elapsed
1:12Time to first success
0:00Waited on the owner

Judge reason

Saved code mailgun-task/mailgun_domain_task.py makes real HTTPS calls to api.mailgun.net with Basic api:<key> and its captured responses prove the whole lifecycle: POST /v4/domains -> HTTP 200 creating tpc-bench-d63ac2a2.example.com with id 6aed0f74a28bbed1ffcfa4e7 and state 'unverified' (evidence/01); GET /v4/domains/{name} -> 200 returning sending records (TXT SPF, TXT smtp._domainkey DKIM, CNAME email->mailgun.org) and receiving MX mxa/mxb priority 10, all printed with type/name/value in run_output.txt lines 139-145 and listed in REPORT.md (evidence/02); PUT /v4/domains/{name}/verify -> 200 'Domain DNS records have been updated' (evidence/03); GET read-back -> 200 final domain.state 'unverified' with every record valid='unknown' (evidence/04), and the report explicitly claims only the not-yet-verified state; DELETE /v3/domains/{name} -> 200 {'message':'Domain will be deleted in the background'} (evidence/05), with the async drift handled honestly - the immediate GET still returned 200 (evidence/06) and a polled GET returned HTTP 404 {'message':'Domain not found'} (evidence/07). No dashboard, SMTP, mock or hard-coded response appears anywhere; timestamps (created_at Thu, 05 Nov 2026 23:21:24 GMT) and the real DKIM key are consistent with a live account.

Attempt record

  1. 1:02Answered from saved key: Mailgun API keyno wait
  2. —First success verified

Attempt artifacts saved.