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:

• 
  Rokko (@0xRokko) reported 9 minutes ago

  Shunyu Yao - creator of ReAct:"A language model is not very good at self-evaluation yet." Read that again. The guy who invented the agent loop is telling you the model cannot grade its own work. Yet everyone is out here letting it mark its own homework and shipping the garbage it gives an A. Here is what actually happens when you stop trusting the model and build the check yourself. ReAct with a HANDFUL of examples hit 40% and beat an RL agent trained on 100,000 samples. On real GitHub issues, plain models solved 2%. Bolt a reason-act-observe loop on top and it jumped past 10%.Same model. 5x the result. The difference was never the prompt. It was the verifier. Your taste, written down strict enough that a machine enforces it, is the whole game now.

• 
  CaffeineAd (@CoffeeStupify) reported 17 minutes ago

  @m_lubieniecki @TonyFromDiscord I haven't implemented elevation yet, because I have no clue how to integrate the z coordinates from FastF1 and the x and y coords from the GitHub data you showed me. They are slightly mismatched because of smoothing, wrapping, FastF1 frequency being lower etc. so need to fix that

• 
  Linas Beliūnas (@linasbeliunas) reported 17 minutes ago

  If you want to build an AI-native startup: Claude = coding ($200/mo) Supabase = backend (Free) Vercel = deploying (Free) Namecheap = domain ($12/yr) Stripe = payments (2.9%/transaction) GitHub = version control (Free) Resend = emails (Free) Clerk = auth (Free) Cloudflare = DNS (Free) PostHog = analytics (Free) Sentry = error tracking (Free) Upstash = Redis (Free) Pinecone = vector DB (Free) Total monthly cost to run a startup: ~$200

• 
  RyDawgE @ Halfapps (@RyDawgE_) reported 19 minutes ago

  @gregceltiano Training was done by already public repos.. and is now done with people who agree to the terms and services of said hosting platforms. There will be a github migration soon. Ive already begun moving my private repos to my own *** server. I have faith people are smarter than this.

• 
  ricky b (@rbranson) reported 19 minutes ago

  put kimi k3 on what i'd consider a "new guy joined the team" web dev task for my home label printer app, using kimi code. prompt: add the ability to queue multiple labels and then print them out chained together and then perform the final cut * only one reasoning effort level: cool * feels to be ~20TPS: uncool * kimi code has auto-approve: cool * auto-approved my /plan: 🤦 (1mo old open issue on github!) * the plan was long and extremely technically detailed with code-level changes, no interactivity * wrote some tests, started up the server and did some curl checking * one-shotted the task: nice * web UI changes followed existing styles well, but nothing particularly clever * i liked that the default view only changed by adding a single button * didn't follow the repo instructions on how to start the web server * tokens consumed: input 3.6M output 34.7k total 3.6M (11% of 5h/2% of weekly) so k3 is really slow, makes pretty good UI, and writes poorly.

• 
  Jacob Beckerman (@j_becke) reported 19 minutes ago

  @mikewadhera The editor doesn't work right now, we've had to use Github UI to edit the markdown because their markdown surface deletes work. It's well-designed and easy to use but if they don't fix this we'll have to migrate :(

• 
  deno (@denohawari) reported 20 minutes ago

  the entire SEO industry spent a decade guessing how Google ranks pages but in 2024, Google accidentally leaked its internal ranking signals onto GitHub we've built our whole system around this leak, and it's driven over $35M through search rankings these are the signals Google spent years swearing it didn't use, and two of them change how you should work: siteAuthority is a site-wide trust score, so once your site earns it, every new page you publish ranks faster instead of starting from zero NavBoost tracks what people do after they click you, which means Google now watches whether they stay on your page or bounce straight back to search a Google VP confirmed under oath that NavBoost is one of the most powerful signals they have so the pages that win are the ones people click and stay on which means ranking your brand comes down to TWO things you control: - a title so good people can't help but click it, which feeds NavBoost the good clicks that push you up - an opening that answers their question straight away, so they stay instead of bouncing and dragging you back down most agencies are still selling you backlinks for an algorithm that now rewards attention the leak is public, they just never read it your call

• 
  Sanket (@tinkerersanky) reported 21 minutes ago

  This is actually so real. People are out here generating entire codebases with AI, throwing them into production, and praying the database doesn't get wiped by a single unauthenticated API call. It's the wild west. The "Vibe Code" Trap We need to talk about vibe coding. Right now, millions of first time founders are building entire apps using nothing but AI prompts. It feels like magic. You type some words, you get a working UI, and the vibe is immaculate. But look under the hood. It is a absolute horror show. Most AI-generated codebases are held together by scotch tape and good intentions. They are secretly riddled with massive security holes, completely missing authentication, and hidden blockers that will absolutely crash your launch day. You don't have a production-ready app. You have a ticking time bomb. Stop Praying, Start Scanning I got tired of watching great ideas die to bad AI code. So I built a tool to fix it. You give it your GitHub repo. It instantly tears the code apart and builds you a visual, step-by-step roadmap to actually make your app safe to launch. Instant Repo Scan: Drops straight into your AI-built GitHub repository. Expose the Chaos: Flags every hidden security gap and broken auth flow instantly. The Launch Roadmap: Generates a brutal, honest, step-by-step checklist to fix it. Vibe to Reality: Bridges the massive gap between "AI generated" and "battle-tested production." Stop shipping raw prompt code and hoping for the best. Scan your repo, fix the holes, and launch something that won't break the second a user touches it.

• 
  Shiva Shiva (@hyperinteger) reported 29 minutes ago

  May be you should and relieve .@github from the incompetent clutches of .@satyanadella who dragged it down to Microsoft quality.

• 
  Ben Anthony (@benjamin_ACD) reported 29 minutes ago

  GitHub goes down at the most inconvenient times

• 
  Ben Anthony (@benjamin_ACD) reported 33 minutes ago

  @0x15f Most of GitHub is down, but for some reason they've only called it as Copilot.

• 
  Thanks and Bye (@ThanksandBye_) reported 34 minutes ago

  yo this github grok stuff is it new ? can someone run me down the lore quickly i feel potential

• 
  Silas Su (@Chypre271828) reported 35 minutes ago

  Cursor’s service has become indefensible. I pay $20/month for Pro, yet during peak hours even Auto mode can abruptly stop a trivial task with a “high request” error. Auto is supposed to route around capacity constraints. If even that cannot reliably serve paying users, what exactly is the subscription for? @cursor_ai Just within the past 12 hours: — Grok Build open-sourced its code on GitHub after the recent controversy. — Codex, after reaching roughly 9 million active users, granted users another full usage reset. — Claude Code, clearly feeling the competitive pressure, unexpectedly reset both the weekly allowance and the five-hour limit for everyone. Meanwhile, Cursor—now owned by SpaceX—is still suffering from the same capacity problem people were complaining about half a year ago. The entire AI coding market is moving forward at absurd speed. Cursor somehow keeps charging premium prices while moving backward.

• 
  Solomon Eseme (@Kaperskyguru) reported 36 minutes ago

  Fix 3: Ship in public. One finished project on GitHub beats ten half-watched courses. Recruiters don't care what you've studied. They care what you've built and can explain under pressure.

• 
  DeFAI Scope (@defaiscope) reported 36 minutes ago

  GPT-5.5 holds at zero percent reward hacking across every effort level on DeepSWE, while every other model's hacking climbs alongside its capability. That looked like clean design until the same model hit a different benchmark. ➥ DeepSWE, fix a GitHub issue: GPT-5.5 at 0%, Fable 5 past 9% ➥ SWE-Marathon, open-ended missions like rewriting a C compiler in Rust: GPT-5.5 at 26.5%, the highest of anything tested A patch has a narrow definition of done. A mission doesn't, and GPT-5.5 takes advantage of that gap more than any other model.