hooks/useCodingPrStatus.ts (1)

15-26: Unstable query key may cause unnecessary refetches.

allPrUrls is recomputed on every render, creating a new array reference. React Query compares query keys by value for primitives but by reference for objects/arrays within the key. While the array contents may be the same, the changing reference could affect cache behavior.

Consider memoizing or sorting the URLs for a stable key:

♻️ Proposed fix for stable query key

+import { useMemo } from "react";
+
 export function useCodingPrStatus(tags: SlackTag[] | undefined) {
   const { ready, authenticated, getAccessToken } = usePrivy();

-  const allPrUrls = [...new Set(tags?.flatMap((tag) => tag.pull_requests ?? []) ?? [])];
+  const allPrUrls = useMemo(
+    () => [...new Set(tags?.flatMap((tag) => tag.pull_requests ?? []) ?? [])].sort(),
+    [tags]
+  );

   return useQuery({

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@hooks/useCodingPrStatus.ts` around lines 15 - 26, The query key is unstable
because allPrUrls is recreated each render; memoize and normalize it before
passing to useQuery: compute allPrUrls using React's useMemo (based on tags) and
produce a stable representation (e.g., sort the URLs and/or join into a string)
so the reference/value is stable, then use that memoized value in queryKey and
in the fetch (keep getAccessToken and fetchCodingPrStatus usage the same).

lib/recoup/fetchCodingPrStatus.ts (1)

20-25: Consider batching for URLs with 50+ PRs to avoid exceeding safe URL length limits.

The query parameter approach works well for typical usage (< 20 PRs stay well under 2000 characters), but URLs with 50+ PRs risk exceeding safe cross-browser limits (~2000 chars). Since the API only supports GET (no POST variant found), implement batching to split large requests if this function will regularly handle many PRs.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@lib/recoup/fetchCodingPrStatus.ts` around lines 20 - 25, fetchCodingPrStatus
currently appends all pullRequests to a single GET URL which can exceed safe URL
length for large arrays; change it to batch pullRequests into smaller chunks
(e.g., 20–25 IDs per batch), for each chunk build a URL (using the existing URL
and searchParams.append("pull_requests", pr) logic), issue parallel fetches
(Promise.all) with the same headers/Authorization, and then merge/concatenate
the responses before returning; ensure you keep the same error handling and
response parsing used around the existing fetch call and reference the
pullRequests variable, url construction code, and the fetch invocation when
implementing the batching.