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:

• 
  magsimich (@magsimich) reported 3 minutes ago

  A MICROSOFT ENGINEER SHOWED AT BUILD 2026 THAT THE WAY YOU HAVE WRITTEN CODE FOR THE LAST 30 YEARS IS BEING REPLACED BY SOMETHING MOST DEVELOPERS HAVE NOT TAKEN SERIOUSLY AND THAT THE TRANSITION ALREADY HAPPENED WITHOUT ASKING YOUR PERMISSION Straight from Microsoft Build 2026 where GitHub Copilot Agent Mode and Copilot Studio Agentic Workflow Builder reached general availability and the phrase intent-first programming stopped being a concept and became a shipping product -> The moment it clicks writing code stops being the job and becomes what it is underneath describing what you want a system to do precisely enough that an agent can build it correctly and audit it safely That one idea reframes everything you thought was your value Why syntax proficiency is no longer the ceiling Why the developer who writes the clearest prompt ships faster than the one who types the cleanest code Why the skill that used to take years to build can now be approximated in seconds by someone who has never opened a terminal Writing code was never the final skill -> writing precise intent that an agent cannot misinterpret is And as agent mode commits rebases and deploys on your machine faster than you can follow the one person who catches the misinterpretation before it hits production is the one who understands what the agent was actually asked to do There is a person on every team who reviews what the agent produced instead of just merging it This is the shift that quietly makes that person irreplaceable Bookmark it The next time an agent ships something broken you will know exactly what question was never asked

• 
  Tejas Vachhani (@imtejasvachhani) reported 7 minutes ago

  GitHub Copilot (AI + Momentum) The physics: Momentum p = m·v — mass (substance of your skill) times velocity (speed of execution). AI acts as a force multiplier on v, but cannot supply m. Application: A developer's mass is their understanding of system architecture, problem logic, and code quality. Velocity is how fast they type and debug. Copilot eliminates the high-friction parts of velocity: boilerplate code, syntax lookup, repetitive patterns. The developer stays in flow state longer, so their velocity increases dramatically. But if a junior dev with no mass (no architectural understanding) uses Copilot to ship code at high velocity, the result is a fragile, buggy system — fast garbage. The winning formula: solid senior developers amplify their existing mass with AI velocity, building momentum that's incredibly hard to stop.

• 
  andrewthecoder (@_andrewthecoder) reported 9 minutes ago

  @JesseStojan yeah the bin stuff won't work on windows, I had that in my head... the .gitignore stuff... yeah! hadn't actually noticed that. thanks. SEE? this is the **** I am trying to get. do you mine creating an issue on github?

• 
  Lama (@lamacodes) reported 12 minutes ago

  @shub0414 Sora is closed.. Perplexity hype is down.. Llama was good at that time since it was open sourced GitHub copilot is blunder... Cursor hype is dead due to claude code..

• 
  Scarab (@ScarabOfficial) reported 13 minutes ago

  I discovered a flag to disable comfy-aimdo thanks to the #GitHub repo' Issues page, so am disabling it for now, but apparently that flag will be removed from the repo' soon. The #ComfyUI people had better get the #LTX23 issue fixed before then.

• 
  Cyfrin Audits (@cyfrin) reported 13 minutes ago

  The manual workflow is brutal. Cross-reference OSV. Check GitHub Security Advisories. Search Socket. Then trace the dependency tree backward to figure out which direct package even introduced it. Most teams don't do this. Not because they're lazy. Because it takes hours per issue.

• 
  Mike Gannotti (@MichaelGannotti) reported 16 minutes ago

  @joshtisdale @Microsoft When it first came up were you presented with two buttons? one to login with Microsoft 365 and One for GitHub?

• 
  Temidaradev (@temidaradev) reported 23 minutes ago

  @halbour727 or you can open issues on github if there is no duplicate but dm is okay

• 
  EdKo (@EdKolife) reported 30 minutes ago

  Google just shrank 31GB of AI memory down to 4GB. Same search. Faster than the industry standard. No training required. This is not a model improvement. This is not a new architecture. → It's a compression algorithm that makes the hardware problem smaller. Right now, running serious AI locally means serious RAM. Most machines can't do it. Most phones can't do it. Most edge devices can't do it. Turbovec quietly changes that math. A 10 million document search engine that used to need a server now fits on a laptop. Nobody is talking about this because it shipped as a GitHub repo, not a press release. The models get the headlines. The infrastructure is where the shift actually happens.

• 
  Vishwas Sharma | DevOps · Security · MLOps (@syssignals) reported 30 minutes ago

  @CaptainInsightX You forgot "expo dependency conflict that requires reading 14 github issues and copying one specific patch from a comment posted in 2024". Round trips on mobile builds in 2026 are still measured in hours and nobody at the tooling companies seems particularly bothered by this.

• 
  Fokki (@0x_fokki) reported 33 minutes ago

  Someone posted a video of a man asleep at a football stadium on a Tuesday. Forty thousand people mocked him before halftime. Fell asleep at the match. Wasted the ticket. Missed the goal. Every sports account shared it by Wednesday. Someone in the replies posted: "respect." His account had one pinned post. A Claude Code terminal. /loop running. Routines active. Auto Mode on. Seven GitHub PRs reviewed while he slept in that seat. Three Slack digests posted. One CI failure triaged, root cause identified, draft fix PR opened. He set up 14 steps of configuration the weekend before. Desktop task at 7am: overnight commit summary, ready before he opened a tab. Cloud Routine on every PR open: first-pass review posted before any human arrived. /loop every 10 minutes: deployment status checked, no one watching. Auto Mode approved 93% of the actions automatically. The people who mocked him watched 90 minutes of football and went home. Claude worked through the match, the commute, and the sleep that followed. He wasn't asleep at the game. He was testing the stack. full 14-step automation guide in the article above👇

• 
  Carver (@carverfomo) reported 34 minutes ago

  A Chinese mathematician posted a 3 minute video on Bilibili explaining how he lost his $10,000 a month gig to AI. The model he had been training started writing harder math problems than he could invent. He admitted his own mistake in business positioning. He had spent four years hand writing PhD level math problems for Scale AI's reinforcement learning pipeline. $50 to $100 per problem. 200 problems a month. Then synthetic data killed his entire contract category. He was no longer able to invent a problem the machine could not solve. At 2:13 he says the word agent. He says it once. He never says it again in the video. The way he says it is the only thing on screen that did not come off the teleprompter. He has been recording videos off a teleprompter for three months. The teleprompter runs on the same agent that killed his Scale AI work. Every script is generated by Claude. Every word he reads to camera is the agent's. The new job is reading. Someone pulled the script repository from a Cursor instance the dev had left public. The folder was labeled bilibili-laments. Inside were 47 video scripts. All in his voice. All written by Claude. Six months ago a 14 year old in Shenzhen pushed an AI agent to GitHub. Judges said no real world application. 3,100 forks later. The mathematician had been one of them. He had wired the agent into his content pipeline the week Scale AI cut him off. He had been a PhD candidate at one of the top five Chinese math schools. He taught there for two years before going full time on Scale AI contracts. He still has the credentials. He still has the office. He just no longer writes anything. He wanted to show people how AI took his career. He accidentally showed them how AI also took his post mortem.

• 
  MICKEYMOUSE (@0x3cc309) reported 35 minutes ago

  Who else is having issues with GitHub premium?

• 
  -TheEscapistЯandom -Baltic Citizen (@theescapistspl1) reported 36 minutes ago

  @github should add new type of error or reason for disabling reposotoy, to be a Compromised Repo!

• 
  nicole (@nicolekcha) reported 37 minutes ago

  Calling this an “outage, not a bug at the inference layer” is a sleight of hand here. It’s still a bug *somewhere* even if not *at the inference layer*. And it’s one of the worst classes of bugs. If an outage can cause your responses to go to the wrong requester, across tenant boundaries, it’s still an *extremely bad* bug. Even if it is due to a bug in a dependency, like a load balancer — you still passed that bug down to your users. You can root cause it and do a post mortem, but it still ultimately is your responsibility. (Like the Vercel data breach through an AI tool an employee used or the supply chain attack on TanStack through a Github Actions bug). Many users reading will ride along with the wording “outage, not a bug in our inference” and not realize “still a bug somewhere in the stack”, taking comfort from a technical distinction that doesn’t earn it. And if that’s what was intended here, that’s a very bad look, and a PR red flag.