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:

• 
  Kevin Tabet (@TabetKevin) reported 7 minutes ago

  @upstash Hey guys i think login with github is broken can't log in rn will try later. google works email i dont have

• 
  John D. Clay (@JohnDClayAuthor) reported 8 minutes ago

  @XFreeze I tried out the new update to Grok Build last night and put it to the test. It helped me go back to a far previous session, it actually has all sessions in a nice area to look at and choose from. I challenged it to fix a broken framework I had built with the earlier versions of Grok Build and with the help of @grok too. I had published it a couple weeks ago and it was not working well. But now after a couple prompts... clayforge the first ai-matove framework for multi agent UI's. You should check it out if you are coding with AI. It's on GitHub.

• 
  Conglomerate (@0xconglomerate) reported 9 minutes ago

  Why exactly do VLAs fail? VLAs start w/ LLMs as their brain. Early roboticists (2021-2022) noticed that LLMs trained on internet text had absorbed a large amount of implicit knowledge about the physical world. So they took that best available pretrained brain, observed that actions could be formatted like language tokens, and assumed the transfer would work. But world knowledge encoded in language ≠ physics simulation. There's essentially a data structure mismatch: ▸ LLM pretraining data is discrete, symbolic, and sequential (text). ▸ Physical control is continuous, high-dimensional, and requires split-second feedback. --- ➦ VLAs in the real world, by the numbers: ① They barely work ▸ VLAs start at ~30% success on real robot tasks, it need hundreds of human interventions just to reach ~90% ▸ Best pretrained VLA hit 27.4% task progress on real robots ② VLAs can't generalize outside training ▸ On actions it's never seen, best VLAs score 25-32% task progress (fails when you change the environment) ③ Fine-tuning doesn't help ▸ The more robot-specific, the dumber it gets at everything else (only works on clean, controlled, success-only demos) ④ Too slow for a real robot ▸ OpenVLA runs at 3-5 Hz (physical control needs orders of magnitude faster than that) --- The easiest way to understand how VLAs are actually wrong is thru a real life example. ➦ Let's say you hired a chef who learned everything about cooking by reading, but has never stepped in a kitchen. If you ask them how to cook a steak, they'll tell you the best answer. But if you actually ask them to cook, they'll struggle when you hand them the pan. They'll have a hard time picking up the ingredients. They'll burn the steak. They know everything about cooking, but can't actually cook. --- ➦ Thoughts I want to take back a line I've said before: "Robots can see, but they still can't listen." (referencing to my Silencio piece before) I take it back. Robots can see, listen, even reason now. What they can't do is act in the real world. It's basically an AI chatbot wrapped in a robot body, not a robot that can actually do tasks. No wonder most demos online are scripted. There's a real problem with the brain, and roboticists have been building on the wrong foundation. VLAs are like a trojan horse, they look like the answer but bring a bunch of problems in with them. VLAs only learn through imitation which brings up the data problem. "Enough data" at scale doesn't mean hundreds of demos total. It means hundreds per task, per robot body, per environment. Hundreds again every time any one of those changes. So you've basically got a human-labor bottleneck. To get that data, someone has to physically collect it, either through: ▸ Teleoperation (slow, expensive, needs trained operators) ▸ Kinesthetic teaching (tedious, doesn't scale to complex tasks) ▸ Motion capture (high precision but high setup cost) ▸ Simulation (robots trained in sim often fail in the real world because physics engines aren't accurate enough) And you'd think, okay, maybe someday a company figures out a better way to collect all this. But the problem doesn't stop once you already have the data... Switch to a new robot body and you're collecting data from scratch, because VLAs don't transfer well across embodiments. Move it to a new environment and you're collecting again, since it just overfits to whatever setup it trained on. Give it a new task and yep, collect again, because it can't generalize to actions it hasn't seen. And if you fine-tune it for one thing, you'll probably break another, so now you're collecting data again just to fix what broke. So what was @DrJimFan and @nvidia's answer to this? World Action Models. Instead of building on a language model, you build on a world model: a model that's learned to simulate how the physical world actually behaves. VLA: a language model that learned to output actions WAM: a world simulator that learned to output actions So when you give a VLA a new task, it needs hundreds of demos to learn it. Give a WAM the same task and it simulates it forward first, acts based on that simulation, then adapts with barely any data. This is what NVIDIA did with the first WAM: DreamZero. DreamZero learns by watching the world (any video of anything, not just robot demos). The backbone is a video diffusion model, the same kind of model that generates realistic video. It was pretrained on massive amounts of internet video, so it already learned how the physical world works: how objects fall, how surfaces interact, how motion flows. Doesn't sound like an entirely different approach, right? But NVIDIA looked at it from a different angle. They figured motor actions are shaped a lot like pixels; both are high-dimensional continuous signals. So DreamZero processes them in the same model, at the same time. It predicts the next video frame and the next action together, through the same architecture. So when a robot runs DreamZero, it's literally dreaming a few seconds into the future in video, then reading its own dream to decide what to do next. If the dream looks coherent, the action works. If the dream hallucinates, the action fails. The DreamZero paper dropped last February 2026, and it's been open source on GitHub for anyone to try. Then in March 2026, at GTC, NVIDIA previewed GR00T N2, the direct successor to DreamZero. This is the production version of the WAM architecture, built for humanoid robots at scale And so far, everything's looking promising. GR00T N2 hits a 98% success rate on unseen domestic objects, a 40% jump over GR00T N1 (the VLA), and 2x better generalization than the leading VLAs. NVIDIA swapped robotics' data problem for a compute problem. Instead of collecting more human demos, just simulate more. So yeah, feels like we're finally pointed in the right direction, closer to robots that can actually function in the real world. Excited to see where DreamZero / GR00T N2 goes from here.

• 
  Alpha Batcher (@alphabatcher) reported 12 minutes ago

  David Soria Parra: "2026 is all about connectivity, and the best agents use every available method" A coding agent needs access to the same places you check while building: - repo and PRs - docs - browser - database - error logs - Figma - tasks - payments The article gives the 11 MCP servers for that setup: - Context7, GitHub, Playwright first - Supabase or Neon, Sentry, Firecrawl next - Figma, Linear, Stripe when you need them - Filesystem, ***, Memory, Sequential Thinking as the base Read it if you keep copying code, docs, schemas, screenshots, errors, and tickets into Claude Code by hand

• 
  AJ ✝️ 💚🧡 (@angelcreative) reported 13 minutes ago

  @uiux_hamad My design team is leaving Figma gradually, in fact we are using Cursor and GitHub as main design tools now, in the past two months the usage of Figma drops 33% and it will keep going down up to 30% more to a 63% in total and maybe more

• 
  Akshay Shinde (@ConsciousRide) reported 17 minutes ago

  @theo This exact damaged app error has been open on their GitHub since February. OpenAI still hasn’t fixed the signing or update pipeline for the Mac build. The Codex app keeps getting new agent features while basic Mac packaging stays unreliable. Priorities are obvious.

• 
  MaxMusterman (@maxschuetz_) reported 19 minutes ago

  New Hack: Tell Codex to search for Github Issues which don't need specific Design Questions. Then say: Spin Up Sessions which Fix each Issue and they use also Subagents. Babysit them until the end.

• 
  Chris Huber (@chubes4) reported 24 minutes ago

  @CoastalDigital2 @MythThrazz That part is more of an idea right now. I need to test it on my VPS. The goal is that non technical users can open issues and PRs against the corresponding live site code on GitHub without touching the production site, safely previewing all changes via Playground.

• 
  Jose (@SolutionsCay) reported 26 minutes ago

  Two changes to how I work with agents: 1. GitHub App so the agents manage issues directly. Keeps the repo clear of throwaway spec and todo files. 2. EmDash (Cloudflare's serverless WordPress successor) for internal docs. Runs on D1, just SQLite under the hood, so I can export the content and move it anywhere. No more docs sprawl.

• 
  Mike Muturi (@_muturimike) reported 26 minutes ago

  Hello @github on 2FA, SMS setup kenya 🇰🇪 is not in the list of countries, is it an error or deliberate omission? Kindly fix it @github @GithubProjects

• 
  Akinsete Motunrayo (@Harkinsete) reported 28 minutes ago

  I built my entire personal brand with AI and a clear process. Here is exactly what I built and how I did it, because you can do this too. What I Built ✅ Brand Strategy (mission, vision, values) ✅ Visual identity: colors, fonts, logo, brand guidelines ✅ A full pitch deck (12 slides) ✅ A speaker kit PDF ✅ A complete multi-page personal brand website ✅ A free lead magnet (a guide people can actually use) How I Built the Website Step 1: I planned before I touched anything I wrote down my brand colors, my fonts, my page structure, and what I wanted each page to do. Most people skip this. Everything breaks when you skip this. Step 2: I gave Claude one detailed prompt with my brand colors, fonts, pages, and copy. It returned a complete, mobile-responsive, multi-page website as a single HTML file. One file. Ready to deploy. The prompt I used: - "Build me a complete personal brand website as a single HTML file. Pages: Home, About, Services, Portfolio, Contact. Primary color [your hex], accent color [your hex], background [your hex]. Display font [font name], body font [font name]. Home page needs: dark hero with my name, photo on the right, tagline, and a CTA button. Services section. Impact numbers. Mobile responsive. No frameworks." Copy this, edit your details, and fine-tune as you want. Step 3: I pushed to GitHub: Free. This took me less than five minutes. Now every update I make is version-controlled and safe. Step 4: I deployed to Vercel for free. Connected my GitHub repo to Vercel and the site was live in under few minutes. This requires no hosting fees and nothing to manage. Step 5: I bought my domain on Namecheap - Searched for my full name and found the .com. Bought it for less than $12 for the year. Added it to Vercel. Updated the DNS settings on Namecheap. Waited 20 minutes. My website was live at my own domain. - Total cost: less than $12. - Total time to go live: under 2 hours. I am also working on a mobile app. A Progressive Web App, which means anyone can visit the URL on their phone and add it to their home screen like a real app. I may be running a live training in July where I will walk you through this entire process step by step to build your live website with a custom domain. If you have a phone and a laptop, you can do this. I documented everything the steps, the exact AI prompts, the domain checklist, the deploy instructions in a free PDF guide. Comment BRAND IDENTITY below and I will send it straight to your inbox. 💾SAVE THIS POST. You will want to come back to it. 🔁 SHARE IT with someone who keeps saying they need a website. The only thing standing between you and a professional online presence is the decision to start. Love and Light, Motunrayo 🤍

• 
  Vishal Tiwari (@VishalTiwa91817) reported 32 minutes ago

  @AlfieJCarter I am a Computer science student . I have given a brief introduction about MCP server in my college and explained them how to connect your GitHub repositories with MCP and your local system with MCP SERVER . I would love to connect you.

• 
  Pipeshub ( Open Source Alternative To Glean ) (@PipesHub) reported 36 minutes ago

  Pipelines are built. Context is broken. MCP is quickly becoming the default interface for enterprise AI agents. And that’s a good thing. It gives agents a standard way to connect with tools and data. Connecting an AI agent to Slack, Jira, GitHub, and Salesforce doesn’t mean it suddenly understands your business. It just means it can access your data silos. In short: "MCP gives your agent a passport. It doesn't give them a map." As enterprise AI undergoes a massive platform shift from passive chatbots to autonomous agentic workflows, this naive, runtime "federated search" approach creates an ugly cycle in production: - The Latency Spike: Slower agent execution while waiting for multiple external APIs to respond before it can even begin reasoning. - The Token Bleed: Skyrocketing bills from shoveling raw, unranked JSON dumps into a massive context window, praying the model finds the answer. - The Governance Nightmare: A massive risk of data leaks if you rely on a base LLM to magically guess and police complex enterprise security permissions on the fly. Agents do not fail because they lack intelligence. They fail because they lack the right enterprise context. The hardest problem in enterprise AI isn't connecting to systems. MCP solved that. The hardest problem is Context Engineering. MCP is the perfect interface, but a permission-aware context layer must be the foundation. 🚀 If AI is becoming core enterprise infrastructure, you cannot allow the strategic intelligence layer of your company to sit inside someone else's managed, closed-box platform. That is exactly why we built Pipeshub (open-source developer owned context infrastructure layer). TL;DR MCP gives agents access. A context layer gives them understanding. And deep understanding is the only way enterprise AI moves from a cool demo to secure, reliable production. 👉 Next Up Tomorrow: MCP Token Tax

• 
  𝕊ℍ𝕀ℕ𝔸☃ (@Shinawritesbugs) reported 36 minutes ago

  @viii_fn Github was slow too

• 
  Pascual ⚡ (@0xPascual) reported 37 minutes ago

  A high school kid opens an account, plugs in Claude 5, and turns a few hundred dollars of lunch money into a six-figure trading account over the weekend. The screenshot goes viral, the replies fill up with people begging for the GitHub repo, and the standard engagement-bait influencers declare the dawn of the sovereign teenage day-trader. The media thought that was the story. It was not. The real flex wasn't the macro strategy or the directional bets on currency pairs. It was the setup behind it: a lightweight proxy array routing through residential IPs to dodge exchange rate-limiting, paired with a custom parsing engine that instantly translates raw order-book imbalances into executed micro-hedges. The kid wasn't trading; he bypassed the entire institutional pipeline of risk management, brokerage compliance, and analyst overhead with a single configuration file. The entire operation runs on a continuous loop of multi-agent orchestration. A master instance drafts the execution logic, a secondary validation agent checks the code against real-time oracle feeds, and a fleet of worker APIs executes up to 3,210 trades a night. Total infrastructure cost: roughly $45 in API tokens and a cheap server instance. It extracts a 78% win rate out of systemic market inefficiencies, operating with a structural margin that legacy trading desks weighed down by salaries and compliance boards cannot compete with.