Benchmark report

Task verified

7m 35s elapsed

The saved attempt met the documented success criterion.

Experience score

89/ 100

Agent-Ready
  • SInstall
  • AAuth
  • SExecute
  • ADocs
  • SCurrent
Experience measures friction in the tested task. The verdict says whether it succeeded.
Agent profile for this run: Install S, Auth A, Execute S, Docs A, Current SSInstallAAuthSExecuteADocsSCurrent
Agent profile · this run
Judge's note · what slowed the agent most
Cloudflare's 403 error 1010 UA ban on us.okx.com blocked the first signed calls until a browser-like User-Agent was added.

Where the time went

7:35 agent time
  • Research0:22
  • Attempt5:13
  • Verify2:00
7m 35sTotal elapsed
3:29Time to first success
0:00Waited on the owner

Judge reason

Saved evidence proves a live us.okx.com BTC-USDC round trip via hand-written stdlib HMAC-signed V5 REST calls (code/okxlib.py; no SDK, headers only). responses/04_orders_pending_before.json shows 0 open orders pre-trade (printed). responses/07+08+09: market buy clOrdId ctrl0mrt1794466185, sz=20 tgtCcy=quote_ccy, state filled, avgPx 73493.1, fillSz 0.00027213 BTC, fee -0.000000952455 BTC. Net 0.000271177545 rounded down to lotSz 1e-8 = 0.00027117, which is exactly the sell size in responses/10-12 (filled, avgPx 73465.7, fee -0.0697259285415 USDC). Printed figures in outputs/console_transcript.txt match the fill JSON exactly, and my recomputation (75 - 19.999677303 + 19.8519679404585 = 74.8522906374585 USDC; residual BTC 7.545e-9) matches the exchange-reported ending balance in responses/13_balance_end.json to the last digit. Net quote loss 0.1477 USDC (<1 USD), residual BTC worth 0.00055 USD (<0.50). orders-history (responses/14) shows only these two orders in the run window (buy 1794466186, sell 1794466219; transcript spans 06:47-06:52), older canceled ctrl0lol/ctrl0mo orders predate it by hours. BTC-USD confirmed nonexistent (code 51001) so USDC is the quote currency. A grep of all saved files found no credential or proxy-token material; credentials were requested as secret fields and read from files outside the artifact tree.

Attempt record

  1. 1:01Answered from saved key: OKX API credentialsno wait
  2. —First success verified

Attempt artifacts saved.