← Back to the carnage
GITHUB

🟠 GitHub decided 1-2% of API requests weren't worth authenticating between 14:49-16:45 UTC. Your Slack/Teams integrations got ghosted. Endpoints played 'who are you?' with legitimate users, serving 4xx errors like denial of service with a smile. (Auto-resolved: no updates in 24h)

⏰ Jun 05, 2026 Fixed (for now) spotted via status_rss

// The official story

Auth issue resulting in API impacts, including some Slack and Teams channel subscriptions

// Status page says

Jun 5 , 20:34 UTC Update - Additional detail on the scope of impact during the 14:49 UTC to 16:45 UTC window: a small but elevated percentage of authenticated requests to GitHub.com received incorrect authorization failures. We saw a 1 to 2% increase in 4xx responses for a small number of endpoints (/repos/{owner}/{repo}/pulls/{pull_number}, /repos/{owner}/{repo}, /repos/{owner}/{repo}/contents/{path}). The vast majority of requests

πŸ”— Official status page (for the PR-approved version)

https://www.githubstatus.com/incidents/2nmfnbknhlnv

Github's Greatest Hits

🟒 GITHUB 6d ago
🟠 GitHub decided to ghost developers twice in one dayβ€”Git ops, Issues, Actions, and PRs all took an unscheduled siesta from 16:52-17:01 UTC. Systems are back online, but the incident playlist is starting to feel like a broken record. (Auto-resolved: no updates in 24h)
🟒 GITHUB Aug 17, 2026
πŸ”΄ GitHub decided to play roulette with your API callsβ€”20% of them lose. Archive downloads? Those are running at 50% failure, which is basically a coin flip. Root cause still a mystery, like finding a specific commit in a monorepo at 2am. (Auto-resolved: no updates in 24h)
🟒 GITHUB Aug 12, 2026
🟑 GitHub's decided logins and asset downloads need a 'thinking break.' Engineers are currently in that special Slack-based panic state they call 'investigating.' Classic move: auth systems remember passwords better than uptime. (Auto-resolved: no updates in 24h)
🟒 GITHUB Aug 12, 2026
🟑 GitHub users reporting degraded logins and package downloads while engineers squint at dashboards pretending this is fine. The 'investigating' status translates to 'we have no idea yet but coffee is being consumed.' At least they're honest about not knowing what broke. (Auto-resolved: no updates in 24h)