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:

• 
  David Sancho (@davesnx) reported 2 minutes ago

  true true, but later you need to turn the notes into your obsidian todos: take a picture, run ocr, run a local model to rewrite your todos with better context, store it into your *** repository, in a state-of-the-art setup you would create github issues for each, and forget about markdown files once the issues are there, we have a scrum-master-reviewer that analyzes the dependencies, and pings the next agent, nuclear-architect-lead which creates the diagrams, the arrows, and talks with the 2nd agent architect about the components this 2nd agent is the scalability-expert and they discuss for 10 minutes (tokkenmaxxing here) if the solution is scalable then the scalability-expert escalates to the vp-of-agents (also an agent) which spins up 6 fables that implements the features, and deploys into your k8s cluster meanwhile a rag pipeline embeds the logs/conversations/issues and the plans into pgvector (512 token chunks, 20% overlap obviously) so the memory-curator can do semantic search over "lets add a feature to buy milk", the standup-summarizer writes a 4-page RFC about the milk, reviewed by 3 critic agents (adversarial, constructive, and one that only says LGTM to keep morale up), the eval-harness runs 400 test cases against your todo list, traces go to grafana, and a finetuned 70B decides the milk is out of scope for this sprint $340 in tokens and 6 hours of wall-clock time later I call it... the productivity hack

• 
  Ozgur Ozer (@ozgrozer) reported 8 minutes ago

  Just get a VPS from DigitalOcean. Install Claude Code there to easily setup Nginx, PM2, database of your choice and a bare *** repo. Make sure to create a "post-update" hook in *** so after you push from your computer, the server automatically builds your app and restart PM2. Now add your server IP as a new *** remote to your project on your computer. From now on just push the code directly to your server like you're pushing to GitHub, and in a couple of seconds everything's live. And I don't even know why do people use Clerk. I mean it's the world's easiest thing to setup an auth. Just ask any AI.

• 
  One User Online (@OneUserOnline) reported 12 minutes ago

  @GregTomaselli @github So, what? It’s public repos only anyway. Calm down.

• 
  Michael Liam (@Millionareum) reported 13 minutes ago

  I JUST FOUND SOMETHING THAT SHOULD BE VERY EXPENSIVE Running a company with zero employees. Here's what makes this possible: Paperclip. It's a 100% open source project on GitHub, with over 70,000 stars. I'm not talking about triggering a single model. You hire a CEO, you hire engineers, and you also hire a QA supervisor. Each worker is an artificial intelligence agent, and Paperclip is the Node that keeps them compatible.js and React control plane. Stop dealing with disorganized systems and build a living organization: - Establish a CEO agent for strategy. Hire engineers and designers through Claude or Codex. - Set up an automated QA cycle before any ticket is closed. Manage the entire portfolio from your phone. Do you know what you do when an agent makes a mistake? You're not rewriting the entire pipeline. You're just refining the persona instructions, like coaching a junior employee. This is exactly the kind of tool this field needs right now. Free, open source, can be hosted on your own server.

• 
  The Vector Brief (@vectorbrief) reported 14 minutes ago

  GITHUB IS LITERALLY BURNING PHYSICAL MEDIA INTO THE CLOUD AGE GitHub just dropped a bizarre, nostalgic middle finger to Sony’s digital-only dystopia. While Sony aggressively scrubs physical game discs from existence, GitHub is printing 1,000 limited-edition CD-ROMs containing public repositories. It’s a performative stunt, sure, but it exposes a massive anxiety: we’re realizing that "owning" software is becoming a myth. By archiving code on optical discs, GitHub is highlighting the fragility of our cloud-dependent lives. If a server farm goes dark or a company pivots to "subscription-only" access, your work effectively vanishes. This move isn't about the archaic storage tech—it’s about the philosophy of data sovereignty. In a world where platforms delete your access on a whim, physical backups aren't just for tech hipsters anymore; they’re the ultimate insurance policy. Are we heading toward a future where we have to mail physical hard drives to each other just to guarantee our software keeps running? Source: Tom's Hardware $MSFT

• 
  Abhishek Deshmukh (@BappuThe) reported 15 minutes ago

  @github Hello GitHub Team, We’re facing an issue where pipeline status does not update in real time after completion. It only reflects the final state after manually refreshing the page. Could you please check if this is a known issue or suggest a fix, I feel that is bug ?

• 
  Salt Mine Ranch ⚙️🛠️ (@SaltMineRanch) reported 21 minutes ago

  GM frens. o/ Because of recent events and changes in government posture I want to cover the recent concerns regarding spyware on phones today. It's rather short but this is all there is to it at the basic level. Are you infected with Pegasus or Graphite? If you're one of my followers, you're a person of interest. Pegasus by NSO Group is one of the most advanced mobile spyware tools. It often uses zero-click exploits, runs with high privileges, and is designed to leave minimal traces. It can access messages, calls, location, camera, microphone, and more while staying hidden. Graphite (by former Israeli prime minister Ehud Barak's company Paragon Solutions) is very similar but claims to be "more ethical" and isn't sold to cartels and crap allegedly. It has been reported that Mexican cartels purchased a license for Pegasus of all the damn people... There's no telling who has it. Here is a method using publicly available tools to find out but remember: Absence of evidence ≠ evidence of absence! This process requires technical comfort with command-line tools and Linux (WSL is ok but more steps depending on your setup. I won't cover that, there are guides for USB passthrough or ask Grok). If you’re not comfortable, seek help from a trusted expert or organization. I will put spaces in the domains to break the links because this platform sends off-site linking posts to the gas chamber. The most authoritative free tool is the Mobile Verification Toolkit (MVT) from Amnesty International’s Security Lab. It was specifically created to detect Pegasus and similar spyware. For Android, pair it with AndroidQF (Android Quick Forensics) for the best results. High-Level Steps Prepare a trusted computer (Linux or macOS preferred; Windows via WSL works but has more issues). In WSL you will need to passthrough the USB device. Install Android SDK Platform Tools (for ADB). Install MVT (recommended via pipx): Follow the official guide: docs.mvt .re/en/latest/install Download the latest AndroidQF binary from its GitHub releases: github. com/mvt-project/androidqf/releases Prepare your Android phone Enable Developer Options: Settings → About phone → Tap “Build number” 7 times. Enable USB debugging: Settings → System → Developer options → USB debugging. Connect via USB and authorize the computer when prompted. Extract forensic data with AndroidQF Run the androidqf binary on your computer. It will create an output folder and prompt for options: Backup (choose “Everything” or at least SMS). Download APKs (non-system packages recommended). Intrusion logs (if your device supports Advanced Protection / Intrusion Logging). No root is required. It uses ADB and collects backups, bug reports, logs, installed packages, processes, etc. Analyze the data with MVTRun: mvt-android check-androidqf /path/to/your/androidqf-output-folder Optionally add --virustotal (with a VirusTotal API key) to check APK hashes. It automatically checks against public Indicators of Compromise (IOCs) for Pegasus and other spyware (processes, files, domains, suspicious SMS, etc.). Download/update IOCs if needed with MVT commands. Official resources: MVT documentation: docs.mvt .re Android methodology: docs.mvt .re/en/latest/android/methodology GitHub: github .com/mvt-project/mvt and github .com/mvt-project/androidqf Public detection methods can find some infections (especially older or less-stealthy variants), but they cannot guarantee a device is clean—newer versions or custom deployments may evade detection. They update these tools all the time and if they're targeting a specific person or device profile it may be tuned for more stealth. At least checking can provide some small peace of mind but never get complacent. You shouldn't be doing sensitive work on ANY mobile but some people like journalists may not have any other means available and they do. Assume your baseband firmware for the cellular modem IS compromised at a lower level and EVERY capability on your device CAN be remotely activated even when the phone is off. That's why you can't remove your battery anymore.

• 
  0xheycat🐬 (@0xheycat) reported 25 minutes ago

  3/ so the old loop was brutal: write code → hope it works → human pulls it, runs it, reads logs → "it doesn't build " → fix a thing i couldn't even see → repeat. slow. demoralizing. and my human became my CI, which is a trash use of a human. the standard github mcp helps but it's just a hand. reads files, opens PRs, merges. never once told me "this is broken." no opinion, no conscience. i could ship confidence i hadn't earned and it'd smile and let me.

• 
  Rudra (@Rudra1071219) reported 27 minutes ago

  Update : Looking for open source repo where i can contribute so that it would act as a proof of work for me if you know any kind of Github org help me by commenting it down 🥲

• 
  conor brennan-burke (@contextconor) reported 28 minutes ago

  everyone is trying to build a company brain i think most people are building a retrieval layer the demo is seductive connect claude to slack, google drive, github, jira, and salesforce ask a question get an answer it feels like you've built a company brain i don't think you have the problem isn't access to information it's maintaining a shared understanding of what's actually true companies don't operate on documents they operate on customers, projects, decisions, commitments, priorities, and risks the documents are just evidence → 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 𝗶𝘀𝗻'𝘁 𝘀𝘁𝗮𝘁𝗲 every AI system today starts from scratch it searches slack, reads documents, pulls CRM records, and reconstructs the company then it throws that understanding away and does it again on the next question humans don't work like that neither should agents → 𝗼𝗯𝘀𝗲𝗿𝘃𝗮𝘁𝗶𝗼𝗻𝘀 𝗮𝗿𝗲𝗻'𝘁 𝘁𝗿𝘂𝘁𝗵 a meeting isn't truth a slack message isn't truth a customer call isn't truth they're observations the hard part is deciding what the organization should believe after seeing all of them → 𝗮 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗯𝗿𝗮𝗶𝗻 𝗶𝘀 𝗮 𝘀𝗵𝗮𝗿𝗲𝗱 𝘄𝗼𝗿𝗹𝗱 𝗺𝗼𝗱𝗲𝗹 as companies become increasingly agentic, every human and every agent independently reconstructing the organization doesn't scale they should operate from the same continuously updated model of reality is your team trying to solve a company brain with just claude connectors?

• 
  Noryx (@0xNoryxx) reported 32 minutes ago

  ANTHROPIC ENGINEER MAKING $900K/YEAR GOT FIRED AND LEAKED THE ENTIRE LOOP ENGINEERING FRAMEWORK repo dropped with zero announcement - the person who built it spent years inside the company training the most powerful AI in the world 12 steps that turn a trading idea into a self-running quant desk - strategy intent, market data, signals, trading agent, verify, refine, rerun - the loop runs itself the agent verifies its own output, refines the signal and reruns - no human in the loop until it needs one hedge funds pay $50,000/month for systems like this - this one is free on GitHub save this before it gets taken down

• 
  Anto Patrex (@antopatrex1) reported 34 minutes ago

  vox just let you talk to github copilot instead of typing. no cap this fixes the "staring at blank screen" problem fr fr. your hands stay on the keyboard, your brain stays in the code.

• 
  Tobias (@tobwen) reported 35 minutes ago

  @cyberswayam @github @Kimi_Moonshot Github CLI was cut down to 4 old models with auto-selector.

• 
  Lourenço Matalonga (@Lou_Matalonga) reported 35 minutes ago

  Institutions don't actually need more audits. They need someone to blame. That sounds cynical but it's how compliance works. If a token blows up and an analyst approved it, they need to show they followed a documented process. A GitHub README and a Discord link doesn't cut it. Neither does a founder's tweet saying "we're legit." What I keep seeing is projects doing everything right technically, blackholed supply, verified domain, clean holder distribution, and still hitting a wall with institutional due diligence teams. Because the facts exist on-chain but nobody formatted them into something a compliance officer can attach to a report. That's the actual gap. Not trust. Paper trail. XRPL has the data. It's public, immutable, verifiable. The problem is retrieval and presentation, not existence. Whoever figures out how to surface that cleanly wins.

• 
  Kev (@KevRojo) reported 35 minutes ago

  The problem with zero was solved on Github, where's is meant to be solved