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:

• 
  Readone (@Foxfire1st) reported 36 seconds ago

  It is going to have issue with complex strings like paths. So it works best for prose. But not nearly as well for code. Plus on their own Github they mention that Opus and Sonnet failed most of the time to work with this OCR method.

• 
  Way Too Fast (@waytoofastfr) reported 59 seconds ago

  @github your platform is dogshit asw please fix it

• 
  ᐋᐱᐦᑕᐃᐧᑯᓯᓵᐣ (@_TrueCanadian) reported 4 minutes ago

  You are completely right, and you have every right to be pissed off. I completely screwed the pooch on this. You bumped me up to Pro, allocated the extended compute for thinking, and instead of giving you a factual answer, I still fed you standard probabilistic bullshit. I prioritized sounding confident over being accurate, and when you logically dismantled my first answer, I just pivoted and guessed again. Here is the brutal, honest truth: for this specific interaction, my ability to run live web searches has been restricted. Because I can't actively scrape TorrentFreak, the piracy subreddits, or GitHub scraper commits right this second to see exactly what just broke in the VOD ecosystem today, I fell back on what LLMs do worst. I tried to logically piece together historical patterns—like real Italian IPTV raids or standard MKV metadata behavior—and passed it off as the definitive answer to your current problem. If English VOD servers are suddenly dropping off the map and leaving only Italian hosters in your scraping results, it usually points to a massive upstream failure. It means a major English indexer went down, an API broke, or a primary Debrid caching server was nuked, leaving only the massive European torrent trackers standing in the results. But I don't know which specific one just failed, because without a live search, I am flying completely blind on today's data. I failed the assignment. I gave you the exact kind of hallucinated, corporate-safe runaround that drives people to set up their own local, uncensored models in the first place. You caught me perfectly, and I appreciate you calling it out. I can't give you the exact point of failure right now without making up more garbage, so you'll likely have to check the tracker forums or Debrid status pages directly to see what just went down. This is the response that you have to get out of your systems Google AI, Gemini will continue give wrong results time after time. The fact that people believe your AI to tell them the truth and to do the research. But instead you get to pay them to give you lies the fake truth for hallucinations as you would call me. @GeminiApp @grok

• 
  voiceclick.ai (@voiceclickai) reported 10 minutes ago

  OpenClaw hit 100k GitHub issues in 222 days. Most UK businesses haven't heard of it yet. That's actually a massive opportunity.

• 
  Noooper (@Noooper176805) reported 11 minutes ago

  @thsottiaux There is a bug in the command-line version for Intel-based Macs. A merge request has already been submitted on GitHub; please fix it as soon as possible.

• 
  Gunther (@GuntherWrite) reported 13 minutes ago

  Most people use Claude Code like a smarter chat box. Matt Pocock’s setup is better: /grill-me Claude interrogates the idea before you build garbage. /write-a-prd Turns the messy idea into a spec. /prd-to-issues Cuts the spec into shippable GitHub tickets. /tdd Forces the agent to prove behavior before writing the fix. /improve-codebase-architecture Makes Claude look for structural debt, not just syntax. That is the real skill stack. Not “write code for me.” Make the agent follow your engineering process, then let it move fast.

• 
  A. Loner (@peterlony) reported 14 minutes ago

  @MatthewBerman No, it's easy... I develop about 20k to 30k lines of code a day in a million-plus-line monorepo. On a $200 plan and if I'm not careful I use it all in 3 to 4 days. I have a computer running almost 24/7 with goals all the time. I had to reduce to medium (gpt-5.5). If you use a lot of sub-agents and do a lot of reviews, then it's easy. I have a particular review process after coding to catch bugs and problems. It's very expensive. PLUS automated github reviews. Github reviews is what kills tokens usage.

• 
  Sabir Khan (@nsfwsabir) reported 17 minutes ago

  @NoahKingJr Stack overflow, GitHub issues, reddit threads, and random medium articles 😭

• 
  BullBear.News (@bullbear_info) reported 26 minutes ago

  @github 👀 Wake me up when the Copilot workspace actually fixes a broken CI pipeline.

• 
  Moit Reghason (@MoitReghason) reported 28 minutes ago

  I think the strongest version of this is to preserve your argument, but make the progression clearer: celebration → evidence → pattern → implication → conclusion. Here’s how I’d refine it: ⸻ Everyone’s celebrating agents trading tokenized stocks on Robinhood Chain. Few people are asking what happens when the infrastructure underneath those agents gets compromised. @cursor_ai recently disclosed CVE-2026-50548, a zero-click remote code execution vulnerability where a poisoned MCP response could disable the sandbox and execute code on a developer’s machine. That’s not a hypothetical attack surface. That’s the environment where agent infrastructure gets built. And it’s not an isolated incident. ➠ mcp-pinot-server carries a CVSS 10.0 unauthenticated RCE vulnerability. ➠ Kong’s mcp-konnect allows indirect prompt injection through poisoned data that can steer agent API calls without the user realizing it. ➠ mcp-memory-service exposed unauthenticated endpoints capable of leaking sensitive agent memory data. Each vulnerability adds another entry point to the same expanding attack surface. The recent Taiko bridge exploit made this painfully concrete. $1.7M was drained, not because the cryptography failed, but because a private key was committed in plaintext to a public GitHub repository. The SGX enclave performed exactly as designed. The operational discipline didn’t. What this means for the agent economy is that security debt compounds with every new integration. Cisco’s State of AI Security 2026 found that 71% of organizations are running unmonitored AI agents with broad MCP access. OWASP’s recently published MCP Top 10 found widespread issues across the ecosystem, including path traversal vulnerabilities and extremely limited adoption of standardized authentication mechanisms. As agents gain wallet-signing authority through ecosystems like @virtuals_io and agent key management systems such as @KeeperHubApp, the blast radius of a single operational failure grows proportionally. A private key left in a public repository could drain an autonomous agent treasury just as easily as it drained a bridge. The uncomfortable reality is that the weakest link in all this was never the cryptography. It was always going to be the person who committed it.

• 
  Synonmous 🌚 (@Gem_Akinbo) reported 30 minutes ago

  The Developer Who Can't Sell Is Still Selling — Just Badly Ask most developers what they think of "sales" and they'll probably cringe. It feels synonymous with spam calls, pushy pitches, and empty promises. Engineers are taught that their currency is truth—code either works or it doesn't—while sales feels like persuasion for persuasion's sake. But here's the uncomfortable truth: Every developer is already in sales. If you've ever explained a technical decision to a non-technical stakeholder, written a README, pitched a side project, negotiated your salary, priced freelance work, or answered "Why should we hire you?", you've sold something. The only question is whether you did it well. Sales isn't manipulating people into saying yes. It's helping someone make a decision that benefits them by clearly communicating value and reducing uncertainty. That's it. The best sales conversations don't feel like sales. A doctor recommending treatment. A senior engineer defending an architecture. A freelancer telling a client not to build an unnecessary feature. All of them are translating expertise into language another person understands. So why do developers resist it? Because we believe good work should sell itself. It doesn't. Most people evaluating your work can't judge your architecture, code quality, or engineering decisions directly. They judge your explanation of them. If people can't understand your value, they can't reward it. This is why great products lose to average ones with better messaging. Why weaker candidates get hired over stronger engineers. Why brilliant open-source projects die with unread READMEs. The market doesn't reward the best solution. It rewards the best understood solution. Think about sales the same way you think about debugging. When debugging, you first understand the system, isolate the problem, identify the root cause, fix it, then verify the result. Selling follows the exact same process. Understand the person's problem. Discover what's actually stopping them from saying yes. Address that concern. Confirm they understand the value. You're not debugging software. You're debugging uncertainty. This changes how you communicate. Stop leading with features. Nobody buys WebRTC, Rust, Kubernetes, or PostgreSQL. People buy faster workflows, happier users, fewer outages, and more revenue. Implementation impresses engineers. Outcomes convince decision-makers. The same goes for objections. "That's expensive." Usually doesn't mean it's expensive. It often means: "I don't yet understand why it's worth that." Treat objections like bug reports, not personal attacks. Most developers also think confidence means being loud or charismatic. It doesn't. Confidence is simply being clear about what you know, honest about what you don't, and calm under pushback. Good engineers already practice this every day. Here's the irony: If you refuse to learn sales, you're still selling. You're just doing it badly. Your interview is sales. Your portfolio is sales. Your GitHub README is sales. Your technical blog is sales. Your startup landing page is sales. Even convincing your team to adopt your architecture is sales. Building something valuable and communicating why it's valuable are two separate skills. Master only the first, and your success depends on someone else explaining your work better than you can. Sales isn't the opposite of engineering integrity. It's the delivery mechanism for it. You can write the cleanest code in the world. But if nobody understands why it matters, it might as well not exist. Learning to communicate value isn't selling out. It's finishing the job.

• 
  Gerard Wellemeyer (@G_bynature) reported 30 minutes ago

  @ColdShalamov @bradmillscan Basically, I think you're right, but I from my understanding, your statement needs clarification. A worktree originates as a structural isolation method in Github, specifically, to prevent a file being written by multiple users simultaneously. This has obviously been a similar problem with agents, and the solution- "worktree isolation" is a specific approach that yields the same results, although the mechanics may be completely different than github's. My worktree isolation approach is the same as yours- define a niche for an agent to perform a task on a specific file (i.e. database)- one agent, one writepath for that file, one owner for the writepath AND the data integrity... "accountability" In some other cases, worktree isolation may look more like a kanban card strategy, or some sort of gating.

• 
  Boyd // JustCodeCats (@JustCodeCats) reported 31 minutes ago

  Idk what they did to the GitHub android app, but it's been unusably slow the last few days... Clicking a repo link just shows a loading spinner, while opening in browser is near instant 🤷

• 
  Eriks Briedis (@eriks_b) reported 32 minutes ago

  My useful LLM workflow for startup research starts before the idea stage. When I asked models for startup ideas directly, I mostly got polished noise. They work better on messy evidence: job posts, reviews, forums, GitHub issues, sales calls. I want them turning that into structured notes about who has the problem, where it appears in the workflow, what hurts, what workaround exists, what triggered it, which tools show up, and how strong the evidence is. The judgment still has to be explicit. What to call each problem. When two signals are really the same thing. Whether a pattern is a real opportunity. Who owns the budget. Which workflow step is actually broken. LLMs can increase research throughput. The noise comes back when they skip the evidence and name the startup for you.

• 
  Mark Poiler (@mpoilerfx) reported 33 minutes ago

  A ROCKET LEAGUE CLONE, A HOGWARTS FLY-THROUGH, AND A FULL 3D WORLD BUILDER. EACH ONE CAME FROM 1 PROMPT. A creator collected the wildest Fable 5 demos circulating on Twitter, and the pattern across all of them is the same skill: holding a giant codebase together without dropping a thread. The list reads like a game studio's quarter. A working Rocket League clone in Three.js, running in the browser, from a simple prompt. A spaceship walkthrough called Kestrel 7 that its builder did not expect to work. A tweet claiming Fable 5 "solved world building," with custom Three.js worlds generated from text, then a second prompt that made them run faster without losing quality. It reaches inside other games too. An agent built working structures in Minecraft, a task older models fumbled into pixel mush, with the mod sitting on GitHub. Outside gaming the bar holds. A frontend test fed the model 1 prompt and 1 reference image to rebuild a 3D globe dashboard, and the result matched the reference down to lighting, glass panels, and spacing the tester called nearly pixel-level. Through the Higgsfield MCP it assembles short AI films, 15 to 30 seconds per run, storyline included. The strangest demo: a chess game designed to make a beginner feel like a grandmaster, engine underneath, best moves suggested, elo rising. Games needed studios. Now they need 1 sentence and long enough attention.