Files
DocsGPT/tests/api/answer
arc53-machine 24796ae61f Mint request ids on the server and trace refused requests
build_agent no longer takes request_id from the request body: it becomes
the primary LLM's usage request id, and quotas count distinct request ids,
so a client could make every call count as one. Requests refused after
setup started (unauthorized, over quota, resume conflict, setup error) now
write their trace, marked error, through an after-request hook; streaming
routes hand the trace to complete_stream instead.
2026-09-23 22:09:58 +01:00
..