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:

• 
  Bprime - Ninjapay (@biellonuhu) reported 6 minutes ago

  Recently I took notice of some happenings. I think we need to talk about access to services as a big concern. My GitHub account was suspended 2 weeks back in the middle of my normal routine, I saw another person tweet about how his OpenAI subscription was suspended in the middle of work, just a few minutes ago I saw another post on how someone's Claude account was suspended as well. In most of these cases it was an automated trigger system that suspended the accounts. It takes forever, if ever, to resolve the issues. No real human looks at it, just some black-box AI deciding your fate in seconds. Now let's look at the impacts on users, very terrible. You lose money, time and opportunities. Deadlines get missed, clients get angry, projects stall, and sometimes you even lose paid subscriptions with no refund. For freelancers and indie developers this can literally mean no income for weeks. I think we need to find a way around these issues. Keep local backups of everything, use multiple accounts where possible, lean more on open-source tools, and push for actual human review systems. Big tech can't keep holding our work hostage like this. This is one example

• 
  prodip (@potu_eth) reported 7 minutes ago

  gm chat. everyone's still asleep on this and that's the actual opportunity quantum computing breaking bitcoin/eth cryptography isn't a "someday" problem anymore. @quipnetwork's own doomsday clock puts real probability on a cryptographically relevant quantum computer by 2028. that's not decades away that's basically tomorrow in crypto timelines what makes quip different from every other "quantum" project larping as research they're not asking you to migrate assets or trust some theoretical roadmap. you wrap your existing setup with a quip firewall contract and get quantum safe encryption without touching what you already hold testnet is live. real users protecting real value right now not a whitepaper promise and it's not just defense. they're building the actual compute layer with classical + quantum nodes solving real optimization problems verified results EVM compatible and open source on github. dwave partnership backing the quantum subnet is not nothing backed by portal ventures and orange dao. team includes people who've actually shipped quantum resistant infra before from the xx network background crypto has a habit of ignoring infra risk until it's an emergency. this is the one category where being early isn't optional it's the whole game quantum won't ask permission. neither should your security @quipnetwork

• 
  Chubigans (@metatransformr) reported 8 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.

• 
  marc.fargas (@marc_fargas_) reported 9 minutes ago

  @theo @zeu_dev I think there's a lot of confusion between the fork/clone buttons in GitHub and what a fork means as a concept. A *** **clone** makes a copy of the code on your computer, *** fork same but server side, A FOSS **fork** takes the code to start a new project with a different team.

• 
  mrgadget (@mrgadgetstudio) reported 11 minutes ago

  @saidotdev Why not working in VSC - github?

• 
  Ajay (@ajay4ai) reported 11 minutes ago

  How I'd become a Forward Deployed Engineer in 2026 if I had to start from scratch. (bookmark this) Most people think AI companies are hiring people to train models. They're not. They're hiring engineers who can take an AI model and make it work inside a real business. That's what a Forward Deployed Engineer (FDE) does. Here's the roadmap I'd follow: 1. Understand the role first. You're not building products for millions of users. You're solving one customer's messy problem at a time. Think of yourself as a founding engineer embedded inside someone else's company. 2. Become broad, not deep. You don't need to master everything. You need to be comfortable switching between: • Backend • Frontend • Cloud • APIs • Databases • AI FDEs win because they connect systems. 3. Learn to ship AI—not train it. Nobody expects you to build GPT-5. Instead, learn: • Prompt Engineering • Model APIs • RAG • Structured Outputs • Evals • Agent Frameworks Production AI beats research every time. 4. Master integrations. The hardest part isn't the model. It's connecting AI with: • Legacy databases • Internal APIs • Authentication • Compliance • Existing workflows This is where most enterprise AI projects fail. 5. Build real AI applications. Forget toy chatbots. Create tools that someone actually uses every day. If nobody depends on your project... ...it's still a demo. 6. Learn MCP and AI agents. Modern AI isn't just prompting. Understand how to build: • MCP Servers • Agent Skills • Multi-agent workflows • Tool calling These are becoming core enterprise AI building blocks. 7. Make AI your coding partner. Use tools like: • Claude Code • Cursor • GitHub Copilot The goal isn't replacing yourself. It's becoming 10× faster. 8. Solve business problems. Customers don't buy LLMs. They buy: • Faster workflows • Lower costs • Higher revenue • Less manual work Always measure success in business outcomes. 9. Practice customer discovery. Before writing code... Ask: • What's broken? • What can't change? • Who uses this? • How is it solved today? The best FDEs spend more time listening than coding. 10. Ship. Maintain. Repeat. Building is only half the job. Real engineering starts after deployment. Fix bugs. Collect feedback. Improve the workflow. That's what companies actually pay for. The AI bottleneck isn't building smarter models anymore. It's finding engineers who can deploy them into messy, real-world environments. That's why Forward Deployed Engineers are becoming one of the highest-paid roles in AI.

• 
  Nikhil Dubey (@nikhild791) reported 13 minutes ago

  @paulg similar answer exists for coding as well we have solution for every github issues code for every question also when it comes to evals we have clear goal of getting which out of x answer is better like leetcode or codechef predict that's why the agents are good at coding

• 
  Bprime - Ninjapay (@biellonuhu) reported 14 minutes ago

  Recently I took notice of some happenings in this space. I think we need to talk about access to essential services as a big concern. My GitHub account was suspended 2 weeks back in the middle of my normal routine, I saw another person tweet about how his OpenAI subscription was suspended in the middle of work, just a few minutes ago I saw another post on how someone's Claude account was suspended as well. In most of these cases it was an automated trigger system that suspended the accounts. It takes forever, if ever, to resolve the issues. No real human looks at it, just some black-box AI deciding your fate in seconds. Now let's look at the impacts on users, very terrible. You lose money, time and opportunities. Deadlines get missed, clients get angry, projects stall, and sometimes you even lose paid subscriptions with no refund. For freelancers and indie developers this can literally mean no income for weeks. I think we need to find a way around these issues. Keep local backups of everything, use multiple accounts where possible, lean more on open-source tools, and push for actual human review systems. Big tech can't keep holding our work hostage like this. This is one example

• 
  Antonio Mele (@antoniomele101) reported 15 minutes ago

  I'm wondering what to make of this post. We already communicate to Codex by voice. we can create agents using Codex as the compute engine on Buzz or Slack or Teams, and Codex is already able to do very long runs on complicated problems. I've done a very long experiment of about 10 days where Codex was asked to first create all the harness (skills, sub agents, plugins, MCPs, GitHub actions, etc) to automate work, and then create something based on a goal. The final outcome was not perfect but it showed the potential. When the models will allow to do this kind of work consistently and reliably, it will be amazing, and my laptop may not even be part of the equation anymore, I would be able to follow the progress on my phone, or ask for voice reports on it. Is this what @thsottiaux is talking about?

• 
  Nitesh (@NiteshTechAI) reported 16 minutes ago

  OpenCut is building a free, open-source video editor that runs in the browser, on desktop, and on mobile. No subscription. No watermark on work you made yourself. • Free and open source, on web, desktop, and mobile • New version built on a single Rust core across platforms • Plugin-first design with real third-party plugins • MCP server so AI agents can drive the editor • Headless mode for batch rendering and automation It is a real answer to the closed editors, and the rebuild adds an editor API and a scripting tab, so you can automate edits instead of clicking every single cut. ⭐ Over 75,000 stars on GitHub. MIT licensed. 🔗 GitHub link in the comments 👇

• 
  MasterMaind .. (@ASaudidos) reported 24 minutes ago

  To everyone who said "he has nothing" here's your firmware version Read it and choke For months a handful of clowns in Discord servers have been running their mouths "He's using ChatGPT" "Anyone can generate that JavaScript" "He's posting bs" "He has nothing" Let me make something very clear The tests you saw before were on firmware 1202 That was the development phase Building the chain Verifying every single step in IDA before touching hardware Not guessing Not copy-pasting from GitHub Not "using ChatGPT to generate JavaScript" There is no JavaScript here Zero This is not a webkit exploit This is not a kernel ROP from 2018 repackaged with a new UI Today the exact same exploit runs on firmware 1352 The output prints the firmware version before anything else so there is no room for your conspiracy theories: PS4 Sandbox Escape - Proof Firmware: 1352 Not 900 Not 1100 1352 Here's what "nothing" looks like: - 389 syscalls unlocked and callable from userland - 60 out of 60 native library functions resolved pipe socket mmap mprotect kill sysctl ioctl kqueue getdents every single one - RAW sockets not just TCP and UDP actual SOCK_RAW on a retail PS4 Go ask your webkit exploit if it can do that - Full filesystem enumeration /dev with 170+ device nodes /app0 /system_ex /system_tmp sandbox internals the disc structure priv directories with DRM licenses cache settings - Live system readout SDK version 1820 CPU core ID 768 MB direct memory console model detection - The sandbox random word (R5JRDruT7f) visible resolved and used to walk the entire sandbox directory tree All of this runs from a BD-J disc on a stock retail PS4 No internet connection required No USB No modification You put the disc in it plays the sandbox dies If you think an LLM can produce code that actually executes on PS4 hardware and returns real kernel data you don't understand what an LLM is what a PS4 is or what exploitation means You understand none of the three To the people who said "look at his past posts on X" yes look at them Then look at this output Then sit down The chain is: Image → Unsafe memory access → SecurityManager kill → Native call patching → Syscall wrappers → Full userland control Every stage proven Every stage running On the latest firmware Sony has shipped to retail consoles This was never about "yapping" This was about building something real while people who have never written a single line of exploit code in their lives sat in Discord calling it fake 1202 was the lab 1352 is the proof One last thing The video attached to this post is running on firmware 1202 that was the development and testing phase A second video is coming soon showing the exact same exploit running on firmware 1352 with full firmware version proof on screen before execution Same chain Same results Latest firmware As for what's next I'm currently in the kernel exploration phase I already have a UAF (Use-After-Free) vulnerability identified in the kernel It's unstable right now and still being thoroughly tested and verified but I'd say it's around 70% there The userland is done The kernel is next Stay tuned or stay quiet And to anyone who claims they've achieved something prove it Show the output Show the firmware version Show the syscalls Show the filesystem Record it on video On a real console On the latest firmware I did It's right here Every stage Every result Your turn Because so far all I've seen from your side is talk Just talk No code No output No proof Nothing but words in a Discord chat I prove my work with evidence You prove yours with arguments We are not the same You're welcome

• 
  ChartsJournalX (@ChartsJournalX) reported 26 minutes ago

  @sameenkarim @github Great, but the main way devs waste time reviewing PRs is GitHub’s clunky, outdated, and slow web UI. Just make it snappy. So much time is lost waiting for things to load.

• 
  Α Ω Programmer (@OmegaNekoSimp) reported 27 minutes ago

  REAL QUESTION: How many of you know that *** isn't just a client to push to GitHub or GitLab? Yall know that *** is the actual server software too and you don't need 3rd parties, right?

• 
  Jakub T. Jankiewicz 🇵🇱 (@jcubic) reported 31 minutes ago

  @piecioshka I use @docusaurus which in comparison, doesn't need a server and can be hosted on GitHub pages.

• 
  nikhil · sys/quests (@nick_realm_01) reported 32 minutes ago

  Read this if you've ever wondered what GitHub Stacked PRs actually solve. I break down: how manual stacked PRs worked why developers had to keep rebasing, pushing, and changing PR bases how GitHub finally made the whole workflow native Should make the whole thing click in under 5 minutes.