Benchmark report

Task verified

4m 6s elapsed

The saved attempt met the documented success criterion.

Experience score

100/ 100

Agent-Ready
  • SInstall
  • SAuth
  • SExecute
  • SDocs
  • SCurrent
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 S, Current SSInstallSAuthSExecuteSDocsSCurrent
Agent profile · this run
Judge's note · what slowed the agent most
Only minor overhead: pulling the OpenAPI spec and stripping the developers HTML to text to confirm the 1-credit price and that enrich_company accepts only a LinkedIn URL.

Where the time went

4:07 agent time
  • Research0:21
  • Attempt2:06
  • Verify1:40
4m 6sTotal elapsed
1:19Time to first success
0:00Waited on the owner

Judge reason

Saved evidence supports every criterion. Docs are saved and correct: docs/openapi.json documents GET /usage ('team-wide credit consumption and current balance') and POST /enrich_company with '**Pricing:** 1 credit per call' and a request schema whose only accepted input is a 'LinkedIn company page URL', so the task-permitted linkedin.com/company/stripe form was correct. code/reconcile_credits.py makes exactly three real HTTPS calls via urllib with the x-api-key header; evidence/run_output.txt shows GET /usage -> HTTP 200, POST /enrich_company -> HTTP 200, GET /usage -> HTTP 200 from a single run (only one script execution in the transcript). Saved responses are genuine: 01_usage_before.json credits_remaining 4965.31 / total_credits 134.69 / enrich_company calls 20, 03_usage_after.json credits_remaining 4964.31 / total_credits 135.69 / enrich_company calls 21, and 02_enrich_company_stripe.json holds real Stripe LinkedIn data (objectUrn 2135371, 16694 staff, request_id req-583e3fd6...) with richapi_meta.credits status 'charged', charged 1.0, remaining 4964.31. Difference of 1.00 credit is computed from the saved responses and matches the documented 1 credit/call, stated as agreeing exactly; the two usage reads are documented free ('Free inquiries, usage, health 0'). Nothing is mocked or hard-coded. No key or token appears in any artifact or log; the key was read from a private file path via the benchmark mechanism and the agent's own grep for the key value over the home dir returned 'no key material in home'.

Attempt record

  1. 1:16Answered from saved key: RichAPI API keyno wait
  2. —First success verified

Attempt artifacts saved.