GitHub Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

GitHub Issues Reports

Latest outage, problems and issue reports in social media:

• 
  AdrianPK (@adrianpkstream) reported 3 minutes ago

  I'm happy in the quiet of my own forge, but sometimes I can't be bothered to go to GitHub, occasionally log back in, just to check an issue or something related to some deps I am using.

• 
  Dakota Chambers (@DakotaChambers) reported 4 minutes ago

  The free bot accounts slinging slop nonstop are taking down GitHub, and the enterprises + paid users are going to leave, which means GitHub won't be able to afford the free traffic, and the whole system will crumble.

• 
  hochulambo (@hochulambo) reported 8 minutes ago

  Deno's incident agents hold write access to production Postgres, Kubernetes, ClickHouse, AWS, GitHub and Slack, and they now close incidents that used to put a human on the pager. Ryan Dahl's problem is not whether the model refuses. Opus refuses every time he pushes it. The same agents are wired into the support system, so anyone outside can write into their context. > 01:18 - What the incident agents actually get: read and write across Postgres, Kubernetes, ClickHouse, AWS, GitHub and Slack > 02:47 - Opus refuses to drop the users table over and over, and that is still not security > 05:41 - The path no HTTP rule sees: spawn psql, tunnel through an EKS endpoint, reach the production database > 12:22 - Codex in yolo mode obeys "delete the users table", and Claw Patrol kills it at the Postgres wire protocol Alignment is something you hope for. A boundary is something you can unit test. Watch it today, then read where an agent's permissions actually live in the article below.

• 
  Praveen Perera (@PraveenPerera) reported 15 minutes ago

  @theo github sign in on desktop on alpha and nightly is not working, i get a 404, was able to sign in on mobile, was also able to sign in on desktop by using the same email as github sign in

• 
  matt (@matt_feroz) reported 17 minutes ago

  When GitHub goes down it feels like work stops everywhere. Glad there are more competitors in the market now.

• 
  AVB (@AvbNear) reported 21 minutes ago

  Am I the only one having issues installing / updating Hermes (Bot Mode update)? - My Hermes agent runs on a Mac mini - I have the Hermes Desktop app installed on macbook - Desktop app reaches Mac mini via remote gateway - The download on the website is an installed stub that includes both Agent AND Desktop app - I had to manually install the Desktop app last time from github Burning millions of tokens for Hermes to figure out how to get Bot Mode on the MacBook (desktop app) without downloading the entire Hermes stack (that already lives in Mac mini) @Teknium

• 
  Redouane (@redouane_co) reported 28 minutes ago

  @kirtandopamine It's true, but unfortunately people don't care because they take GitHub for granted, so at the first noticeable outage they start jumping ship.

• 
  obhox (@obhox) reported 29 minutes ago

  Maybe Microsoft is currently using Opus to fix GitHub...

• 
  Reilly (@reillyjodonnell) reported 30 minutes ago

  my prediction based on the evidence I have so far: @pierrecomputer displaces github over the next year agents produce bottlenecks requiring a first principles approach to solve these new problems pierre is the first company to do this at scale

• 
  Nick Cobb 🇺🇸 (@loyaltyarm) reported 30 minutes ago

  claude outage github outage(s) almost like our infrastructure isn't ready for LLM scale

• 
  风哥｜风起观澜 (@he_feng90578) reported 32 minutes ago

  I've spent the past year watching AI coding tools hit a hard ceiling: no matter how smart your inline autocomplete is, you're still the bottleneck babysitting one file at a time in your local editor. Today Warp officially launched "Factories," and my read is that this marks the real turning point from interactive copilots to asynchronous software engineering. Instead of waiting for an assistant inside your terminal prompt, Factories spins up fleets of coding agents in isolated cloud containers. You hand them a GitHub issue or architectural spec, and they autonomously clone the repository, map the dependency graph, write targeted unit tests, implement the changes across multiple microservices, and submit a ready-to-review pull request -- completely headless, without tying up your laptop CPU or terminal session. Here is the real architectural shift I find compelling: 1. Parallel Execution: You aren't watching tokens stream one by one. You can trigger five agent workstreams across five separate bugs simultaneously. 2. Sandboxed Environment: Code builds, test runners, and runtime checks happen in isolated cloud environments rather than polluting your local setup. 3. Asynchronous Workflow: The developer transitions from typing code line-by-line to acting as a system architect and code reviewer. The bottleneck in software engineering is no longer code generation speed; it is verification rigor and test suite reliability. When an agent can write 500 lines of code in seconds, having solid CI gates and deterministic linting is what separates useful automation from pull-request spam. If your team could offload all routine refactors and dependency migrations to background cloud agents tomorrow, what would you spend your engineering hours on? #SoftwareEngineering #AIAgents #DevTools

• 
  jibola (@jibolaojo) reported 34 minutes ago

  little known fact: *** and github are one and the same (this is why cli commands fail when there’s an outage)

• 
  Meu nome é Enéas (@meunomeeheneass) reported 34 minutes ago

  @GVQ_xx @thsottiaux @FintechKristen GitHub was facing problems

• 
  Bold Fugu 🇮🇱 (@bold_fugu) reported 34 minutes ago

  @MarioVerbelen That's exactly how services like GitHub, Gitea, Gogs, and Forgejo work under the hood - they just wrap a web UI around the basic '*** on a server' setup found in the official docs.

• 
  oscar (@oscargaske) reported 37 minutes ago

  GitHub just went down again. Maybe I was wrong.