mirror of
https://github.com/zed-industries/zed.git
synced 2026-08-24 00:16:18 +00:00
67a5a78682
The ChatGPT Codex backend and the OpenAI API both report an exhausted usage allowance as HTTP 429, with a body distinguishing it from an ordinary short-lived rate limit (usage_limit_reached, usage_not_included, insufficient_quota). from_http_status previously mapped every 429 to RateLimitExceeded, so the agent retried requests that could not succeed until the usage window reset, then surfaced a generic rate limit message. Parse those bodies into a new UsageLimitReached completion error carrying the provider's message and the reset delay, skip automatic retries for it, and render a dedicated callout in the agent panel. |
||
|---|---|---|
| .. | ||
| src | ||
| Cargo.toml | ||
| LICENSE-GPL | ||