GitHub is down right now. Not a slow day. Not a minor hiccup. A full-blown incident hitting web, API, Actions, Copilot, and authentication — all at once.
We tracked the entire thing live. Here’s what happened, why it keeps happening, and what you should do about it.
Live Update: 10:46 pm Indian Standard Time (IST) :- Update 1
GitHub is having an incident and push events to Cloudflare are operating in a degraded state. Check their status page for more details.
What Broke Today? (Timeline)
The first alert fired at 13:40 UTC. GitHub’s status page moved fast this time — within six minutes, half their services were flagged.
Here’s the minute-by-minute breakdown:
| Time (UTC) | What Happened |
|---|---|
| 13:40 | GitHub starts investigating “impacted performance” |
| 13:41 | API Requests marked degraded |
| 13:42 | Actions degraded |
| 13:44 | Webhooks degraded |
| 13:45 | Pull Requests degraded |
| 13:46 | ~20% error rate across PRs, Issues, and more |
| 13:58 | Issues degraded |
| 14:04 | Pull Requests degraded (confirmed) |
| 14:24 | 20% error rate on web + API. 50% error rate on archive/raw downloads |
| 14:31 | Copilot dead. Major outage declared |
That’s 51 minutes from first alert to Copilot going dark.
The Damage Right Now
As of 14:31 UTC, here’s the live status:
| Service | Status |
|---|---|
| Copilot | 🔴 Major Outage |
| API Requests | 🟠 Degraded (~20% errors) |
| Webhooks | 🟠 Degraded |
| Issues | 🟠 Degraded |
| Pull Requests | 🟠 Degraded |
| Actions | 🟠 Degraded |
| SAML/OIDC/SCIM | 🟠 Impacted |
| Archive Downloads | 🔴 ~50% error rate |
| Raw Content | 🔴 ~50% error rate |
| Git Operations | 🟢 Operational |
| Packages | 🟢 Operational |
| Pages | 🟢 Operational |
| Codespaces | 🟢 Operational |
Half of all raw file downloads are failing. If your build system pulls from GitHub raw URLs, half your pipelines just broke.
This Isn’t New — GitHub Has a 2026 Problem
Today’s outage isn’t an isolated event. It’s a pattern.
The Numbers Are Ugly
| Period | Uptime | Industry Standard |
|---|---|---|
| Feb 2026 | ~87% | 99.9% (three nines) |
| Aug 6, 2026 | 10+ hour outage | — |
| Aug 17, 2026 | Ongoing | — |
| 90-day avg (Git Ops) | 99.99% | — |
| 90-day avg (Actions) | 99.43% | — |
February 2026 was below one nine of availability. That’s not a typo. An 87% uptime number for the world’s largest code hosting platform is staggering.
One Hacker News user tracked it: “Github continuing their recent trend of less-than-one 9 of availability.”
Recent Outage History
- Feb 9, 2026 — Five separate incidents in 12 hours. PRs, Actions, Copilot, Webhooks, Pages — all hit.
- Apr 28, 2026 — Degraded performance across issues, packages, searches, workflow runs.
- Aug 6, 2026 — Actions and Pages down for 10+ hours. Webhooks throttled. Jobs queued indefinitely. Copilot review and the coding agent both dead.
- Aug 13, 2026 — Webhooks incident (1h 5m) + GHEC Team Sync disruption (2h 5m).
- Aug 17, 2026 — Today. The one you’re living through.
That’s roughly one major incident every two weeks in 2026.
Why Does GitHub Keep Breaking?
The Azure Migration
GitHub has been migrating from its legacy datacenter to Microsoft Azure since October 2025. The half-migrated state is a known source of instability. Services straddling two infrastructures create complex failure modes that are hard to predict and harder to debug.
AI Broke the Load Curve
This is the 2026 angle nobody talks about enough.
AI-assisted development didn’t just add Copilot to GitHub. It changed how much developers interact with the platform:
- AI coding agents (Copilot Coding Agent, Claude Code, Codex) clone repos, read files, and hit the API constantly
- Developers push more code because AI writes it faster
- CI/CD pipelines run more frequently because more code = more builds
- Webhooks fire more often because more events happen
As one Hacker News commenter put it: “It’s crumbling under the load of a significantly large amount of AI-enabled coding. Users are raising PRs and pushing content a lot more than previously.”
GitHub’s infrastructure hasn’t scaled to match. The demand curve bent upward. The capacity didn’t follow.
The Dependency Chain Is Terrifying
When GitHub goes down in 2026, it’s not just “we can’t push code.” Here’s what actually breaks:
- CI/CD stops — GitHub Actions queues pile up, deploys don’t run
- Webhooks go silent — Slack notifications, monitoring alerts, deploy triggers — all dead
- Go builds fail globally —
go buildpulls from GitHub. Every Go project breaks, not just yours - AI coding tools die — Copilot, Claude Code, Codex all need GitHub for repo context
- Pages sites go down — Docs, marketing sites, dev portals
- Compliance evidence disappears — SOC 2 teams relying on PR history for audit trails lose access
One provider’s bad day becomes everyone’s bad day.
What Developers Are Saying
Hacker News and Reddit lit up the moment things started failing.
The frustration is real:
“Another day, another Github outage. Files are not loading, cannot create repos.”
“I’m wondering how many millions in developer productivity are being lost every minute.”
“If I was a business customer paying for GitHub, I would be very upset with the availability lately.”
“Right at the precise moment I decide to git pull!”
The status page denial is a recurring theme:
“As of the time of writing, nothing on the status page either.”
“Actions are failing and starting new actions results in a 500 error. The status page is shockingly all green.”
GitHub’s status page has historically lagged behind actual outages. Today they were faster — updates started within a minute. But the trust deficit from past incidents remains.
Some are eyeing the exit:
“GitLab is starting to look even more appealing as the 9s disappear from GitHub’s uptime numbers.”
Though, worth noting — GitLab’s own status page showed issues today too.
What You Should Do Right Now
If You’re Affected Today
- Don’t panic-push. Wait for the status page to clear. Force-pushing during outages creates merge conflicts when things recover.
- Check your Actions queue. Jobs may be stuck, not failed. They might resume when services recover.
- Switch to SSH if HTTPS is failing. Git Operations are marked operational — SSH clones/pushes might still work.
- Use cached dependencies. If your builds pull from GitHub raw URLs, switch to local caches or mirrors temporarily.
- Monitor the incident page directly: githubstatus.com/incidents/zkxwbgr0cnmx
Long-Term: Stop Being Single-Source
The real lesson from 2026’s GitHub outages: stop treating any single platform as irreplaceable infrastructure.
- Mirror your repos. GitLab, Gitea, or even a self-hosted Git instance. Keep a hot backup.
- Use dependency caching. Don’t pull from GitHub raw URLs in production builds. Use Artifactory, Nexus, or even a simple S3 bucket.
- Decouple your CI/CD. If your only CI pipeline is GitHub Actions, you have one point of failure. Add a fallback.
- Have a webhook dead-letter queue. If webhooks are critical to your workflow, buffer them. Don’t assume delivery.
- Monitor third-party status pages. Tools like IncidentHub, Updog, or StatusGator aggregate status data and alert you before GitHub’s own page updates.
The Bigger Picture: Is GitHub Becoming a Risk?
This is the question engineering leaders are quietly asking.
GitHub hosts the world’s code. It runs the world’s CI/CD. It powers the world’s AI coding tools. And in 2026, it can’t stay up for two weeks straight.
The platform is trying to do everything — code hosting, project management, CI/CD, code review, AI coding, package registry, static hosting, security scanning — on infrastructure that’s mid-migration and buckling under AI-era load.
The math doesn’t work. You can’t layer more services on top of infrastructure that can’t reliably serve basic operations like pull requests.
For now, GitHub remains the default. But “default” isn’t the same as “dependable.” And every outage pushes teams closer to exploring alternatives.
Frequently Asked Questions
Is GitHub down right now?
Yes. As of August 17, 2026 at 14:31 UTC, GitHub is experiencing a major incident. API requests, Actions, Webhooks, Issues, Pull Requests, and SAML/OIDC authentication are degraded. Copilot is in a full major outage. Archive and raw content downloads have a ~50% error rate.
Why does GitHub keep going down in 2026?
Three factors: (1) an ongoing migration from legacy infrastructure to Azure since October 2025, (2) dramatically increased load from AI coding tools that constantly hit the API, and (3) the platform’s expansion into too many services without proportional infrastructure scaling.
Can I still use Git during the GitHub outage?
Git Operations are currently marked as operational. SSH-based clones and pushes may still work even when the web interface and API are failing. Try switching from HTTPS to SSH.
What’s the alternative to GitHub?
GitLab, Bitbucket, and self-hosted options like Gitea or Forgejo are the main alternatives. However, GitLab also experienced issues today. The real answer is redundancy — mirror your repos across multiple platforms.
How much uptime has GitHub had in 2026?
In February 2026, community trackers recorded ~87% uptime — below one nine of availability. The 90-day rolling average for Actions is 99.43%, while Git Operations sit at 99.99%. Actions reliability is the weakest link.
Last updated: August 17, 2026, 14:31 UTC. Source: githubstatus.com, Hacker News, Reddit, IncidentHub.
