[Feature Request] Review false-positive API refusal for legitimate agent research use case
Bug Description
Appealing a false-positive refusal. This session has been refused since
Oct 3 with category frontier_llm, citing Anthropic's Terms of Service
restrictions on "developing competing model products."
Actual use: internal agent research — skill-library distillation and
automated experiment scheduling on the ALFWorld open-source benchmark,
using open-source models. We do not train, distill from, or benchmark
Claude, and we are not building any model product for external release.
This session ran normally for over four months (opened May 22) with no
refusals before Oct 3, and the nature of the work did not change.
Please review and advise whether this work can continue.
Refused request: req_011CfeVSK8ssdDNDcKqTJQtx
Environment Info
- Platform: linux
- Terminal: vscode
- Version: 2.1.288
- Feedback ID: 8bd17754-87cf-42ef-8160-16e101fa5006