fix: handle false positive Claude AI usage limit errors #8405
Conversation
![roomote[bot]](https://avatars.githubusercontent.com/in/1546624?s=60&v=4){kind=small}
- Add error handling to detect and properly handle malformed rate limit errors - Check for unrealistic future timestamps in error messages - Provide clearer error messages for both real and false rate limits - Add comprehensive test coverage for error handling scenarios Fixes #8404
dosubot
bot
added
size:XL
hannesrudolph
added
the
Issue/PR - Triage
| @@ -0,0 +1 @@ | |||
| Subproject commit 8111da66bd59ca8d500e5eae23b24a0419ed7345 | |||
There was a problem hiding this comment.
P0: Accidental submodule/file under tmp/. This should not be committed. Please remove from the PR and add the path to .gitignore (or exclude via tooling) to prevent reappearance.
| @@ -0,0 +1 @@ | |||
| Subproject commit 7b7bb49572975c4aeff2381a0ebea99b3aa4542c | |||
There was a problem hiding this comment.
P0: Accidental submodule/file under tmp/. Please remove from the PR and ensure tmp paths are ignored.
| } | ||
| } | ||
|
|
||
| // Check for actual rate limit errors from Anthropic API |
There was a problem hiding this comment.
P1: The 429 rate-limit check is inside the 'message is string' guard. If message is undefined/non-string, real 429s won’t be caught. Move this check outside the message-type guard (or add a second guardless check) so it always triggers.
| } | ||
|
|
||
| // Check for actual rate limit errors from Anthropic API | ||
| if (error.status === 429 || error.message.includes("rate_limit_error")) { |
There was a problem hiding this comment.
P1: Same as above for completePrompt(): ensure 429 is handled even when error.message isn’t a string.
| ) | ||
|
|
||
| // Throw a more informative error | ||
| throw new Error( |
There was a problem hiding this comment.
P2: Preserve the original error context. Consider using new Error(msg, { cause: error }) when rethrowing for better debugging.
| // Check for the specific malformed error message pattern | ||
| if (error.message.includes("Claude AI usage limit reached")) { | ||
| // Parse the timestamp if present | ||
| const timestampMatch = error.message.match(/\|(\d+)/) |
There was a problem hiding this comment.
P2: Timestamps may arrive in milliseconds (13 digits). Normalize by dividing by 1000 when the captured string length > 10 before comparing to oneYearFromNow.
| // Check for the specific malformed error message pattern | ||
| if (error.message.includes("Claude AI usage limit reached")) { | ||
| // Parse the timestamp if present | ||
| const timestampMatch = error.message.match(/\|(\d+)/) |
There was a problem hiding this comment.
P2: Same normalization in completePrompt().
|
|
||
| if (timestamp && timestamp > oneYearFromNow) { | ||
| // This is likely a false positive error | ||
| console.error( |
There was a problem hiding this comment.
P3: Prefer the project’s logger over console.error for consistency.
| @@ -0,0 +1,218 @@ | |||
| import { describe, it, expect, vi, beforeEach } from "vitest" | |||
| import { AnthropicHandler } from "../anthropic" | |||
| import { Anthropic } from "@anthropic-ai/sdk" | |||
There was a problem hiding this comment.
P3: 'Anthropic' import appears unused and may fail strict lint rules. Remove or use.
| it("should handle 'Claude AI usage limit reached' without timestamp", async () => { | ||
| const errorMessage = "Claude AI usage limit reached" | ||
|
|
||
| mockClient.messages.create.mockRejectedValue(new Error(errorMessage)) |
There was a problem hiding this comment.
P3: Consider adding tests for (1) millisecond timestamp normalization and (2) status=429 where message is not a string to cover the suggested changes.
Description
This PR addresses Issue #8404 where users are experiencing false positive "Claude AI usage limit reached" errors with unrealistic future timestamps when using Claude Opus 4.1.
Problem
After the Claude Sonnet 4.5 implementation, users started seeing error messages like:
The timestamp (1759770000) represents a date in 2025, which suggests this is a malformed error message rather than a legitimate rate limit.
Solution
Added robust error handling in the Anthropic provider to:
Changes
createMessageandcompletePromptmethodsTesting
Review Confidence
Code review completed with 92% confidence score. Implementation follows existing patterns and includes comprehensive test coverage.
Fixes #8404
Important
Improves error handling in
anthropic.tsto address false positive usage limit errors by checking for future timestamps and malformed messages, with comprehensive tests added.createMessageandcompletePromptinanthropic.tsto detect and handle malformed rate limit errors.anthropic-error-handling.spec.tswith tests for false positive errors, legitimate rate limit errors, and other edge cases.This description was created by
for 1466cc9. You can customize this summary. It will automatically update as commits are pushed.