feat: analysis runs for all fetched issues in parallel with no concurrency limit — for large repos (500+ issues), this fires hundreds of simultaneous OpenRouter API requests and exhausts rate limits within seconds, causing cascading 429 errors

[divyanshim27]

🐛 Problem Statement

IssueScope fetches issues and then performs AI analysis on each one. The README states: "Fetches issue details and performs AI analysis in parallel for high throughput." However, unbounded parallelism is the opposite of high throughput when the downstream API has rate limits.

OpenRouter free-tier rate limits are typically 10-20 requests/minute per key. For a repo with 100 open issues, IssueScope fires 100 concurrent requests, hitting the rate limit within the first second. React Query retries with exponential backoff — but all 100 are retrying simultaneously, creating a thundering herd that ensures the rate limit is hit on every retry wave too.

Result: All 100 issues show "Analysis Failed" with error 429 Too Many Requests. The tool is functionally unusable on any repo with more than ~15 issues.

Proposed Fix

Implement a concurrency-limited analysis queue using p-limit:

// src/hooks/useAnalysis.ts
import pLimit from 'p-limit'; // add to package.json

const CONCURRENT_ANALYSIS_LIMIT = 3; // Conservative — respects free-tier limits

export const useAnalyzeIssues = (issues: GitHubIssue[], apiKey: string) => {
  return useQuery({
    queryKey: ['analysis', issues.map(i => i.number)],
    queryFn: async () => {
      const limit = pLimit(CONCURRENT_ANALYSIS_LIMIT);
      const results = await Promise.all(
        issues.map(issue =>
          limit(async () => {
            try {
              return await analyzeIssue(issue, apiKey);
            } catch (err) {
              if (isRateLimitError(err)) {
                // Exponential backoff with jitter before retrying
                await sleep(2000 + Math.random() * 1000);
                return analyzeIssue(issue, apiKey);
              }
              return { issueNumber: issue.number, error: 'Analysis failed', score: 0 };
            }
          })
        )
      );
      return results;
    },
    enabled: issues.length > 0 && !!apiKey,
  });
};

const isRateLimitError = (err: unknown) =>
  err instanceof Error && err.message.includes('429');

const sleep = (ms: number) => new Promise(resolve => setTimeout(resolve, ms));

Also expose a settings control in the UI for users with paid OpenRouter plans to increase the concurrency limit.

Files to Modify

File	Change
package.json	Add p-limit dependency
src/hooks/useAnalysis.ts	Wrap analysis calls with pLimit(3)
src/components/SettingsPanel.tsx	Add "Concurrency" slider (1–10) for advanced users
src/store/settingsStore.ts	Persist concurrency preference in Zustand
README.md	Document rate limit behaviour and recommended settings

Suggested labels: bug, performance, api, ux

I would like to work on this. Could you please assign it to me?

[divyanshim27]

Hey @vee1e ,
Please assign it to me under GSSoC 2026.
/claim /assign /claim-assignment