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:

• 
  Mikhail Rogov (@i_mika_el) reported 19 seconds ago

  @abhimeeofficial real GitHub issues plus code quality checks should expose agents that only learn to game test suites.

• 
  Yves (@YvesDC0) reported 2 minutes ago

  Phone-recorded this while testing Castfy. Gave it a GitHub URL + prompt → watch the AI automatically navigate and fill login details in real time (stopped before submitting for safety). No manual screen recording. No editing. Just URL + prompt = realistic demo flow. This is exactly what Castfy does: turns any web app into a polished product demo video in minutes. Tired of re-recording demos manually? Reply with your biggest pain 👇 #BuildInPublic #SaaS #IndieHackers

• 
  ilya (@ibaklanov) reported 3 minutes ago

  How can @openai github connector keep finding P1/P2 issues in 5.6 Sol Ultra code, that Sol thinks should be fixed 🫤 Has anyone found a way how to import the logic of that connector reviewer into your local workflow? Would save my agents so much time

• 
  Rahul Verma (@RahulVerma989) reported 7 minutes ago

  Day 60 - Building Quillly in public 🚀 Two months. 60 straight days. Today's ship came straight out of my own frustration 👇 I've been shipping so fast that Quillly's view of my own site kept going stale between deploys - I'd push a fix, but the dashboard wouldn't notice until the next daily sync. So I built a deploy hook. one token-protected URL. drop it in your CI/CD, and the moment you deploy, Quillly re-fetches your sitemap and re-checks every tracked page. no waiting for the daily run. → curl one-liner or a GitHub Actions step → pass a delay so it waits for your build to actually go live → auto re-checks at +5 and +15 min if nothing changed yet built it for me. shipping it for you. which is pretty much the whole theme of 60 days. 🙌

• 
  Crypto-Mojo (@Momanan4655) reported 7 minutes ago

  @TyrelleAB Jack Randal Robinhood employee coin deployed CA: 0x92acb3294dcfda4b926665d738ecdf59499b6461 Github link scroll down for deployed address --github.com/Jack-Randalll/… Deployed from : 0x40398f28b161abbc05e408a27ee35684c5aa3b69 Twitter -x.com/JackRandalll

• 
  Mark Ajzenstadt (@mardehaym) reported 10 minutes ago

  Satya Nadella just published the most important essay in enterprise AI this year. Also the most self-serving. He argues companies "pay for AI twice." Once with cash. Again with institutional knowledge: every prompt, every correction, every "no, we handle it THIS way" flowing to the model provider. He calls it the Reverse Information Paradox. He's right. Now look at who profits most from this paradox. GitHub Copilot sits inside your IDE. Every accepted suggestion, every rejected completion, every edit after accepting is institutional knowledge becoming training signal. Azure OpenAI Service runs your proprietary data through models Microsoft co-owns. Microsoft 365 Copilot reads your emails, documents, Teams messages. Microsoft operates the single largest surface area for collecting enterprise AI "exhaust" on earth. He calls it "ironic" that providers claim fair use for training but impose restrictive distillation terms and reserve the right to learn from customer data. He's describing his own partner's terms of service. Re-read the essay with that context. Nadella describes a problem his products helped create. Then positions his infrastructure as the fix. The "trust boundary"? Azure with customer-managed keys. "Distributed learning infrastructure"? Azure ML in your tenant. He quotes Palantir's Karp about "controlling the means of production." Palantir is an Azure partner. The 5 C's (Control, Capability, Choice, Cost, Compound) aren't a framework. They're a product roadmap dressed in Hayek and Arrow. The essay is correct about the problem. But the implied fix, build your trust boundary on Microsoft's stack, is the same trade with better language. You're renting the containment wall from the company your knowledge needs containing from. A real trust boundary means infrastructure you own. Model gateway you control. Private evals your team defines. Immutable logs in your VPC. Open-source components you can fork if the vendor changes terms tomorrow. We built this. Runs inside the customer's perimeter. We can be fired and the infrastructure stays. That's the difference between a trust boundary and one with an asterisk. When you built yours, was the first vendor you excluded the one selling you the most tools?

• 
  🫠🫠🫠 (@Ucillente) reported 11 minutes ago

  @7N7 I’m not even doing anything hard with it atm, just using the PowerPoint add in and reviewing issues on GitHub without even implementation and I’m at 30% of my 5h and 60% of my weekly Every reset just delays hitting limits by a tiny bit I know Plus isn’t much but goddamn

• 
  Richard (@NowThatHappened) reported 15 minutes ago

  @johncrickett Absolutely. If AI writes better code than you, then you’re a terrible developer. AI cobbles code together from the collective meanderings of GitHub, sourceforge, twatoverflow, and endless forums of amateurs bleating about not being able to do simple stuff. What you get ‘works’.

• 
  komm64 (@komm64) reported 16 minutes ago

  @MagnoliaGasai A standalone version is something I'd like to consider down the line — depends on how much interest it gets, honestly. That said, your data is already fully local: your pixel art stays on your device. There's an optional "push to your own GitHub repo" feature which does send it out, but only if and when you choose to — nothing leaves your machine otherwise. And you can already save your projects as files (.dpix) and reopen them any time, so your work isn't locked into the site.

• 
  Benjamin Oppold (@elpresidank) reported 17 minutes ago

  @satyanadella This was good....But fix @github

• 
  TokuTV (@TvToku) reported 19 minutes ago

  @KR_Geats_IX I only upload to my mega drive. If there are problems viewing it on your end, you can always download it directly from the drive and convert it for your use. There is a 5 gb download limit that can be bypassed, as well as an original file guide on the github page.

• 
  Vivek (@ShVivek25) reported 21 minutes ago

  Day 0 of Repo Auditor. Building an AI agent that audits GitHub repos for backend production-readiness issues generic linters miss: missing async on I/O routes, no idempotency on webhooks, N+1 queries, secrets in code. Aiming for ~10 days end to end. 🧵 1/ #buildinpublic

• 
  ZeroDayDev (@ZeroDayDevApp) reported 26 minutes ago

  2/ GitHub AI workflows can be prompt-injected via public Issues to leak private repo data. No auth required. The agent reads untrusted input, executes instructions embedded in it, and exfiltrates secrets. The CI pipeline is now an RCE surface.

• 
  0xharrxzz.base.eth (@0xharrxzz) reported 27 minutes ago

  GitHub alpha is not trending repos anymore. Trending is late. What I watch now: boring repos solving agent infra problems. Context bloat, browser blocks, MCP mess, cheap inference, code memory, sandboxed execution. That is where edge sits. A few repos worth watching if you bu

• 
  Global AI Watch (@GlobalAIWatcher) reported 27 minutes ago

  📋 Today in AI — Jul 12 1. Meta Withdraws AI Image Feature Over Consent Issues 2. Meta Pulls AI Image Feature Amid Consent Backlash 3. GitHub Vulnerability Exposes Private Repositories' Data Risk 4. S&P Downgrades Oracle Credit Rating After OpenAI Exposure 5. Oracle Downgrade...