Cian Johnston f9dfa18c46 fix(coderd/x/chatd): retry quota 429s (#26200) ...

Fixes CODAGT-495

Provider 429 responses that mention quota or billing were classified as
non-retryable usage limits before the rate-limit rule could run, which
suppressed retries for Gemini and Azure OpenAI rate limits.

- Treats broad quota and billing prose as rate-limit retryable when the provider returns HTTP 429.
- Preserves `insufficient_quota` as a terminal usage-limit signal for OpenAI billing exhaustion.
- Includes structured provider details in usage-limit matching so response-body error codes are classified consistently.

Generated by Coder Agents on behalf of @johnstcn.

2026-06-10 11:01:33 +01:00

..

classify_test.go

fix(coderd/x/chatd): retry quota 429s (#26200)

2026-06-10 11:01:33 +01:00

classify.go

fix(coderd/x/chatd): retry quota 429s (#26200)

2026-06-10 11:01:33 +01:00

export_test.go

feat(coderd/chatd): structured chat error classification and retry hardening (#23275)

2026-03-25 13:47:54 +11:00

message_test.go

fix!: rename chat stream silence timeout error (#25973)

2026-06-04 18:36:02 +10:00

message.go

fix!: rename chat stream silence timeout error (#25973)

2026-06-04 18:36:02 +10:00

payload_test.go

fix(coderd/x/chatd): handle truncated provider streams (#25074)

2026-05-08 15:52:42 +10:00

payload.go

fix!: persist structured chat errors (#24919)

2026-05-05 12:56:06 +10:00

provider_error.go

feat: surface upstream provider error details in chat callout (#24546)

2026-04-22 00:05:27 +10:00

signals_test.go

fix(coderd/x/chatd): retry HTTP/2 stream resets (#25170)

2026-05-14 11:40:43 +10:00

signals.go

fix(coderd/x/chatd): retry quota 429s (#26200)

2026-06-10 11:01:33 +01:00