A caller could change source text, output fields or model policy during asynchronous token counting or execution. That let a response be validated against different evidence or looser limits than the request used. Snapshot request, evidence and policy before the first await and freeze tokenizer messages.
Regression tests reproduce both failures before the fix. Validation: focused tests, typecheck, targeted lint, full suite (1289 passed / 194 skipped), and Next build. The document workflow remains internal and does not prove semantic completeness, provider authorization or production readiness.
A caller could change source text, output fields or model policy during asynchronous token counting or execution. That let a response be validated against different evidence or looser limits than the request used. Snapshot request, evidence and policy before the first await and freeze tokenizer messages.
Regression tests reproduce both failures before the fix. Validation: focused tests, typecheck, targeted lint, full suite (1289 passed / 194 skipped), and Next build. The document workflow remains internal and does not prove semantic completeness, provider authorization or production readiness.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
A caller could change source text, output fields or model policy during asynchronous token counting or execution. That let a response be validated against different evidence or looser limits than the request used. Snapshot request, evidence and policy before the first await and freeze tokenizer messages.
Regression tests reproduce both failures before the fix. Validation: focused tests, typecheck, targeted lint, full suite (1289 passed / 194 skipped), and Next build. The document workflow remains internal and does not prove semantic completeness, provider authorization or production readiness.