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:

• 
  Brian Money 🌻 (@brianmoney) reported 6 minutes ago

  Try to get some work done and @GitHub is down 😢

• 
  Parsa Sabetz (@parsasabetz) reported 9 minutes ago

  @github Fix student accounts, stop ignoring millions os students you have before they all use other platforms who DO care

• 
  Joe mother (@Joemoth11098375) reported 10 minutes ago

  @nai_sucks Then stop letting people use github as a place to download apps and softwares, the problem only exists because the creator deems it so

• 
  betraidx (@betraidx) reported 12 minutes ago

  Every AI you use has amnesia. It just bills you for it. Upload a PDF. Claude reads it. Great answer. Close the tab. Tomorrow it reads the same PDF again. Same tokens. Same cost. Zero memory. You're paying for the same work twice. Every single day. Karpathy looked at this in April 2026 and said: not a model problem. An architecture problem. He fixed it in 800 words. Raw documents are source code. Compile them once into wiki pages. Query the wiki forever. Never touch the raw files again. Three folders. raw/ for originals. wiki/ for compiled knowledge. instructions/ for processing rules. No database. No cloud. One PDF becomes 8-15 interlinked pages. 70-90% fewer tokens on every repeat query. Compounding weekly. That Obsidian graph on screen is the vault after months. Every node a note. Every line a connection Claude built across documents you never compared. 5,000 GitHub stars in days. 41,000 developers followed. Turns out most of us were using AI at 5% and paying for 100%.

• 
  Teja Karlapudi (@teja2495) reported 12 minutes ago

  Quick Search latest stats 🔥 - 33000+ downloads - 1400+ reviews with average rating of 4.8/5 - 123,000+ lines of code - 252 Github issues closed

• 
  kaneryu (@kaneryu) reported 13 minutes ago

  @killbunny_ @V33V33V33V33V33 @CryptoCyberia you're not the intended user for GitHub. This is more of an issue of developers linking to a developer only platform and expecting normal users to know how to use it

• 
  Nav Toor (@heynavtoor) reported 14 minutes ago

  Your phone needs WiFi to send a file to another phone. Or Bluetooth. Or AirDrop, which only works between Apple devices. Or Nearby Share, which only works between Android devices. Or a cable. Or a cloud service that uploads your file to a server in Virginia before sending it to a phone 3 feet away. Every method needs infrastructure. A developer who goes by Alstroph built a file transfer that needs nothing. One phone flashes QR codes on its screen. The other phone points its camera at the screen and reads them. The file reconstructs on the receiving phone. No WiFi. No Bluetooth. No internet. No app. No pairing. No account. No permissions beyond camera access. Just light. It is called Decimen Optical Transfer. Trending on GitHub. Tom's Hardware covered it yesterday. Here is the clever part. The sender never sends the same frame twice. Every frame is a random mix of pieces of the file. The receiver just needs to catch enough frames from any point in the stream and it can mathematically reconstruct the whole file. Missed a frame? Does not matter. Blurry frame? Does not matter. Out of order? Does not matter. This is called a fountain code. Same math used in satellite communication. Roughly 129 KB/s in the demo. Up to 186 KB/s propped steady. A 2 MB image in about 15 seconds. When this matters: Two phones on different networks. Two phones with no cell signal. Two phones in airplane mode. A secure environment where no wireless is allowed but a screen and camera exist. A country where the internet is shut down but screens still glow. One honest note. This is a proof of concept, built overnight with Claude Code. Prior art goes back to divan's txqr in 2018. What makes this version notable is fountain codes running in a browser with zero install, on any phone with a camera. 2,602 stars in 2 days. 309 forks. MIT license. Just a screen. A camera. And light. 100% Open Source. (Link in the comments)

• 
  Jack Paar Fan Account (@wiideenjoydiet) reported 15 minutes ago

  @patrician_tv which is fine, because anybody with a third of a brain cell will have no issues navigating github

• 
  Milady (@onchainmilady) reported 16 minutes ago

  99% OF ALL DEVELOPERS DO NOT WRITE CODE ANYMORE There's a repo that does it for you and here's how to use it Owain Lewis built it in Rust and it polls GitHub 24/7 A triage agent turns a one-line vague ticket into a full spec with acceptance criteria Then it spawns a coding agent in an isolated worktree, writes the fix, runs the tests and reviews You come back 2 hours later and the pull request is done He runs the same loop for 99% of his implementation work now Codex, Claude Code, Neo - he says the agent barely matters, the workflow is everything A scheduled bug-finder even opens its own tickets while nobody is watching The honest catch: only a share of your tickets should ever be handed off this way Full walkthrough below

• 
  Yume_X (@yume_arasaki) reported 16 minutes ago

  @__tinygrad__ Your AIs are rapidly becoming an extension of your mind. The opening volley was that it wasn’t possible to do this locally. That moat is fading away fast. It’s just a matter of time the closed US Labs realize their real moat is UX and harnesses and people will pay them for it. When hardware and optimizations and cost economics all collide, owning your own intelligence for individuals and companies is going to become a topic, ironically it’s exactly the same mindset that drives American closed LLMs to do the same thing. No company wants its trade secrets “uploaded to the cloud” to be trained so a competitor emerges from their hard work The Chinese and the rest of the world are not fully “open source”, they just do a weird dance switching open and closed, it wins them massive loyalty and clout. Today I replaced my main driver with just two sparks. For less than the price of a new car - $8k vs $20-$30k . You can basically completely be AI self sufficient for most things. Yes, it’s a down payment and we should be honest that , AI hardware is going to get better so hardware is just going to depreciate, the break even maths for hardware is brutal vs renting at the moment I was a big fan of ChatGPT 4 , they decommissioned the model. Now I can’t ever be cut off. I have full control. I replaced it with Qwen 3.6 27b Dense , i can fine tune the models exactly how i want them and i don’t need to have it stuck inside some company. I can run the whole thing off-line. I don’t need to be paranoid about whether they honor the “butron” about not using your data. Btw guys if you forgot , GitHub by default has some setting where its on for sending your repo for “improvements”, switch it off if you don’t want your repos uploaded to Microsoft. If you have very proprietary stuff you shouldn’t even be using GitHub. You should host your own repo if you don’t want to secure it. The issue with “training” is unlike before the legal liability is difficult as it will be impossible for you to prove a product was made with your input. It’s the same issue as how artists have no way of claiming that models used their data to create some art that absorbed their style.

• 
  Carl010010 (@carl010010) reported 18 minutes ago

  @davis7 I had to find out about this in the GitHub issues last month. So odd that it’s not on by default.

• 
  Chubigans (@metatransformr) reported 19 minutes ago

  Just ran my first fully autonomous overnight build for my complex Godot RPG with Codex. I used a dispatcher that maintains a concurrency of N=2 (can tune it up as tokens permit) substantial spec-driven tasks at a time. Uses an automated playtesting API to prove the tasks are done. It's easy to make a simple game, it's really hard to make a complex game (mine is Mount & Blade + Wizards) Not everything can be done in an automated way, not even close. But a lot of systems-y tasks, bugfixes and cleanups *can* be done, or at least prototyped relatively autonomously. The workflow is: nightshift runs the backlog off my github project, I wake up and review everything, playtest the build, record bugs and plan out the development for the day. This is spec-driven development for games, and it's definitely the future. The bottleneck is graphics/animation/gamefeel/juice and such. There are so many problems that can't be currently done in an autonomous way, without wasting tokens. This is the "research problem" I (and other AI game engine startups that are vc-backed) am working on. A fascinating field.

• 
  Mahito (@_M4hito) reported 21 minutes ago

  @fuma_nama I just cloned the Fumadocs openAPI version, and I am having a problem where none of the buttons or dropdowns are working at all. I looked for a Discord server, but it seems like you don’t have one. I also got confused by the GitHub discussion so many threads, gosh

• 
  therapistFinder (@therapistFind3r) reported 21 minutes ago

  @Slypador He's only half right. This only started becoming an issue when devs started using GitHub as a program downloading website as well as a development website.

• 
  Zygard (@zygard77) reported 22 minutes ago

  @sflorimm Right now, there are many free tools out there especially in GitHub that you could plug to Claude (that's what I use rn) to get it working in an impressive manner. There are lots of stuff to reduce the tiring thereby saving time. The only issue is that I aint learning anything🥲