Hi team,
I’m reporting a credit-burn issue where agent tasks failed repeatedly without producing usable output, consuming my entire monthly quota.
What happened
Over the past several days I’ve been working with the K3 agent on a document-processing pipeline (OCR → data extraction → SQLite/Metro2 formatting). The agent tasks consistently errored out, produced incorrect outputs, or got stuck in loops requiring me to resubmit the same requests multiple times. Despite the failures, credits were deducted each time.
Specific issues encountered
-
OCR on scanned PDFs returned fabricated data (e.g., incorrect SSNs, mixed-up DOB fields) despite clear source documents
-
File-watcher and automation scripts failed to execute correctly, requiring repeated restarts and corrections
-
Agent repeatedly lost context of prior steps in the workflow, causing circular troubleshooting instead of progress
-
Multiple file-processing tasks returned errors or partial outputs that had to be re-run
Result My monthly credit quota is now fully exhausted and I have no completed workflow to show for it. The bulk of the consumption came from re-trying failed agent tasks, not from successful usage.
Account Info
-
Organization ID:
org-5cc1ab751b774ad1b5ede84c8ecc0d73 -
Creator ID:
d9paq0boF8js9hkt94e0
Request
Please review my account’s recent agent task logs for the failure pattern described above. I’m requesting a credit refund or quota reset so I can continue testing once the underlying reliability issues are resolved.
Happy to provide further details via DM or email if needed.
Thanks.