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:

• 
  Jared Tate ©️ (@jaredctate) reported 1 minute ago

  I'm a blockchain developer. North Korean hackers tried to steal my crypto & hack my computer by pretending to recruit me on LinkedIn for a crypto consulting gig. Here's the entire attack: PHASE 1 — THE LURE A "recruiter" (a polished, good looking blonde "British woman in London") messages you on LinkedIn. Real-looking profile, perfect English. She books a Google Meet. On the call it's a different person — a bald Chinese looking guy, who quickly turns his camera off complaining of a bad connection, broken English — who asks two telling questions: "What crypto wallets do you use?" and "What code editor do you use?" That's not an interview. That's target profiling. PHASE 2 — THE DELIVERY He sends a GitHub repo (a "Web3 game" project) and says: clone it, run `npm install`, and open it in Cursor so we can see you know what your doing. It looks like a normal codebase. It's bait — with malware hidden in 3 separate files, each rigged to run automatically. PHASE 3 — PAYLOAD #1: package.json (the trip wire) They added one line: "prepare": "node server/server.js" npm runs the "prepare" script automatically right after `npm install`. So just installing the dependencies launches their code. No button. No warning. PHASE 4 — PAYLOAD #2: auth.js (the backdoor) In one file, the real code ends at line 18. On line 19 — shoved far off-screen behind ~600 blank spaces so you'd never scroll to it — sits obfuscated code that: • collects your hostname, OS, and MAC address • sends your entire environment (process.env) to a North Korean server: 147.124.212.180:1224 • runs whatever JavaScript that server sends back, via eval() = full remote control • repeats every 5 seconds That's a backdoor. The eval() is the doorway they use to push the actual wallet/password stealer. PHASE 5 — PAYLOAD #3: .vscode/tasks.json (the silent one) A second hidden file set to "runOn": "folderOpen" — it executes the instant you open the folder in your editor, with no window and no output: Translation: download a script from their server and run it immediately. THIS is why they demanded Cursor — they were betting it would auto-run the task. VS Code stops and asks "do you trust the authors of this folder?" first. I never said yes, so it never fired. DO NOT USE CURSOR (for many reasons) PHASE 6 — THE FALLBACK When the repo "wouldn't run," he sent a slick website with a "Connect Wallet" button — a straight-up wallet drainer — and said "just connect MetaMask." THE GOAL Drain your crypto and steal your credentials — money that funds the regime. It's a documented operation: the "Contagious Interview" campaign (aka Famous Chollima). HOW TO NOT GET HIT • A "job interview" that requires running their repo = malware until proven otherwise. • Never `npm install` a stranger's code on your real machine. Use a throwaway VM. • Keep serious crypto in cold storage — never a hot wallet on your daily driver. • If a recruiter's story falls apart the second you get on a video call, walk.

• 
  Geoff Langenderfer (@geoff_l) reported 2 minutes ago

  @petergyang @GergelyOrosz it's evaluating llm performance on known github issues you have a github issue with an attached code change. The llm makes an attempt and you compare it to the known good code diff.

• 
  Sherlocke Bones (@ReturntotheK) reported 6 minutes ago

  @StormWarningMom @AwakenedOutlaw The very serious problem with drones is they really aren't that hard or expensive to build or purchase for basic function. You could learn from YouTube or GitHub or even drone parts manufacturer docs. You don't even have to do any real building beyond cut and paste and bolt together if you don't want to buy one. For wpnization, I wouldn't want to even speculate to give some ahole ideas, but if you can make a drone motor spin, you know enough to control any other basic mechanism. Not that they didn't use AI and maybe Anthropic ignored flags, but they wouldn't need to use AI. That's why drones should be treated with the utmost seriousness and urgency. Pandora's box has been opened, further than the Ukraine war opened it, that's one thing we can definitively determine.

• 
  🅰️nzceel base.eth (@anzceel) reported 6 minutes ago

  Most crypto projects die twice. First when users leave. Then when the lessons disappear. My startup idea is FailureDB. The problem: Every cycle creates thousands of dead DAOs, NFT collections, tokens, and apps. New builders keep repeating the same mistakes because the evidence is scattered across old Discords, governance forums, GitHub repos, and forgotten posts. The solution: An AI-powered protocol that continuously archives failed projects and turns them into searchable intelligence. Ask it: "Why did similar NFT projects fail?" "What warning signs appeared before user growth collapsed?" "Which token models consistently broke down?" Instead of studying only winners, builders could learn from thousands of documented failures. Crypto has become excellent at preserving transactions. We're still terrible at preserving lessons. I'd love to see experiments like this emerge through ecosystems like @RallyOnChain. What failed project taught you the most?

• 
  VAIBHAV TUPE (@VAIBHAVTUPE) reported 7 minutes ago

  ICP KILLS AWS BILLING MODEL FOREVER. Blockchain cloud hosting just went mainstream and devs are sleeping on it. - Internet Computer Protocol runs full-stack apps 100 percent on-chain since 2021. No AWS, no Google Cloud, no centralized servers at all. - ICP trades at $2.40 to $2.60 with $1.34 billion market cap as of June 17, 2026. It ranks #57 to #60 among all cryptocurrencies. - Data egress on ICP costs 300 times less than AWS. But ICP charges for data ingress while AWS gives free inbound data. Storage is cheaper but replicated across all nodes. - Late 2025 upgrades boosted compute throughput by 50 percent and doubled subnet storage to 2 TiB per subnet. Total network storage hit 94 TiB across 47 subnets. - Mission 70 plan approved April 2026 cuts ICP inflation from 9.72 percent down to 2.92 percent to 5.42 percent by end of 2026. Token jumped 39 percent with volume spiking from $70 million to $700 million in 24 hours. - ICP hit 1 billion transactions in Q1 2026. GitHub had 3,196 commits with over 100 active contributors for 9 months straight. DFINITY signed Pakistan for 1,500 AI licenses in Feb 2026. - Canisters act as both code and database with reverse gas model so users pay zero fees. Cloud Engines launched May 2026 for 5th anniversary to target enterprise and gov workloads. 👉 Bottomline : ICP is not crypto hype. It is AWS without Amazon.

• 
  outloop (@0utloop) reported 11 minutes ago

  I had a GitHub repo with my ex. I still create issues, but there's no response anymore.

• 
  katekyy (@snowmews) reported 14 minutes ago

  @Uwu123231634 @nodo519809 @oxcrowx Well then that person can just ignore the hate and delete the issues/PRs. If the author of that repo wouldn't want anyone else to see it or his GitHub then they could either leave it on their computer or make it private.

• 
  shmidt (@shmidtqq) reported 15 minutes ago

  Creator of Claude Code: "Right now you still need to know how to code. In a year or two, it won't matter. I haven't edited a single line by hand since November." In a 90-minute podcast, Boris Cherny breaks down the exact setup behind the tool now writing 4% of every public commit on GitHub. More value than a $500 vibe-coding course. Save this. In a year we'll know if he was right.

• 
  SM TECH (@SMTECHYT2) reported 18 minutes ago

  How do Chinese devs get unlimited Claude code and Codex for free? Yes, really, it 100% works till now. You can set this up in exactly five minutes. > Stop wasting 20-100 dollars every single month on restricted ChatGPT / Claude code Plus plans. > Chinese developers are using a loophole to run autonomous local agents completely free. > Download the official OpenAI Codex / Claude code installer and login. > Download the cc-switch free installer for Windows or your preferred operating system. > I will drop the link in the comments; the GitHub repo has 100k+ stars already. > Use OpenRouter or NVIDIA to get free, unlimited APIs of advanced models. > Open the CC SWITCH app, go to Settings, then Routing, then click Local Routing. > Turn on "Show Routing Toggle on Main Page" and the "Routing Master Switch." > Save settings, go back, and click the Codex or Claude Code agent tab. > Click Add, choose your provider like OpenRouter NVIDIA or Minimax or Kimi or Custom, and paste your free key. > Save it, enable the config, and click the start routing button next to the switch. > Open Codex or Claude Code to see custom models like DeepSeek, MiniMax, or Kimi. ENJOY UNLIMITED CODING WITH CLAUDE CODE OR CODEX OR ANTIGRAVITY OR OPENCLAW OR HERMES. bookmark this post

• 
  Eugene Ostroukhov (@eeuoss) reported 19 minutes ago

  @neogoose_btw Sure. *** has flaws. And let’s separate ****** GitHub UI from inherent *** issues. But: 1. *** was and is a huge improvement over what was before it (CVS, SVN, ClearCase). Effortless local version control is sick! 2. *** is designed for systems programming and works really great for that. Small self-contained changes. No sprawling rewrites.

• 
  Todd Desiato (@TokenDepotCorp) reported 19 minutes ago

  Would you be interested in helping promote real Kaspa adoption? I spent the past year building Oma Wallet, a Kaspa wallet designed specifically for token utility. It is live now, but it launched quietly and did not get much media attention. Oma supports Kaspa, KRC-20 tokens, Issue-Mode CA tokens, offers, swaps, rewards, discounts, subscriptions, memberships, and other real-world token use cases. I am also building AMEKAS, pronounced Am-eh-KAS, like "Am-eh-ri-ca + Kas" with emphasis on KAS. It is a Kaspa and KRC-20 checkout shopping center where sellers can set up online stores that accept Kaspa and approved KRC-20 discount or entitlement tokens at checkout. No dollar checkout, no payment processor fees, and no broker fees. There is also a node operator angle: anyone running a Kaspa node can install a small script that lets them manage subscribers for Oma Wallet and future AMEKAS shopping access. That can turn a Kaspa node into a subscription business tied directly to Kaspa utility. You can learn more on the Token Depot website and GitHub. I know your goal is to promote Kaspa. Would you be willing to take a look and help spread the word?

• 
  Keisuke (@KeisukeIshikawa) reported 19 minutes ago

  CLAUDE FABLE 5 SYSTEM PROMPT AND ANTHROPIC CAN'T DELETE IT. THE GOVERNMENT PULLED THE MODEL. THE INSTRUCTIONS ARE STILL PUBLIC. 120,000 characters. 1,685 lines. 27,000 tokens. every rule anthropic gave the most powerful model it ever shipped sitting in a github repo with 40,000 stars what the leak actually reveals: → fable 5 and mythos 5 are the same model. the only difference is the safety filter layer on top → a persistent storage API most users never knew existed artifacts that save data across sessions → exactly how the copyright limits work: hard 15-word quote cap, one quote per source, then it paraphrases → the full child-safety instruction block, word for word → how it decides when to refuse vs comply, written out in plain english this is the closest thing to seeing behind the curtain that exists. if you ever wondered why claude talks the way it does, or refuses what it refuses, the answer isn't a mystery anymore. it's a text file the repo is run by pliny — the same guy who's been jailbreaking frontier models the day they launch for two years. CL4R1T4S now holds the leaked system prompts for chatgpt, gemini, grok, perplexity, cursor, and every model that matters the original post hit 700,000 views in 48 hours. the repo gained thousands of stars in a weekend here's the part anthropic can't fix: the government can force a model offline in three hours. it cannot un-leak 120,000 characters that already live on 40,000 forked machines the system prompt is the actual product. the weights are the engine, but the prompt is the steering wheel. and the steering wheel is now open source whether anthropic likes it or not we are watching the last secret layer of frontier AI get pried open one leak at a time follow and bookmark before the next wave figures it out

• 
  deci (@realdecimalist) reported 22 minutes ago

  no cardano isn’t cursed. its just struggling with the same brutal reality most smaller ecosystems face.. low liquidity, low activity, and a chicken-and-egg problem for new protocols. taptools pausing/winding down operations is legitimately bad news and i’m not going to try to sugarcoat that ****. it was THE go-to analytics and portfolio tool for cardano defi, serving over a million users. reasons were leadership exodus (multiple co-founders, CTO, COO, etc.) + unsustainable operating costs in a low-activity ecosystem. @IOHK_Charles even warned publicly that more projects would collapse this year due to the tough market and governance challenges. this is a symptom, not the root cause. the bitter pill to swallow: defi tvl: $96 million (down recently). top protocols like minswap, liqwid, etc. make up most of it. solana by comparison: ~$4.9 billion tvl roughly 50x higher. daily active addresses are hovering around 12k–28k recently (spikes happen, but baseline is modest). dex volume is obscenely low (~$2.3M in 24h). chain fees: tiny (~$1.3k in 24h). cardano has strong fundamentals, high staking ratio (often 58-70%), solid security, formal verification roots, and decent github activity. but on-chain usage and capital deployment are nowhere near the hyped chains. alot of holders are long-term stakers rather than active defi degens. why is adoption so low? cardano’s “research-first, slow and steady” philosophy worked great for building a secure base layer, but it hurt in the speed game. competitors (solana, base, newer L2s) iterate faster, have better dev UX for many, and attract more hype/marketing/liquidity. low tvl creates a vicious cycle.. new protocols have trouble bootstrapping liquidity and users. governance friction (recent failed votes on funding, like the summit). broader crypto market has been tough. attention flows to memes, high-velocity trading, and chains with explosive narratives. is it hopeless? not necessarily. cardano has survived multiple cycles with a dedicated (often frustrated) community and patient capital. real-world use cases like parametric insurance, RWAs, or enterprise stuff could be where it differentiates long-term, rather than trying to out-meme solana. but the ecosystem does need more activity to support dApps. taptools dying is a warning shot that cardano businesses are under pressure. most defi protocols die or stay tiny. the ones that succeed usually do it in bull markets with strong product-market fit & distribution. cardano isn’t dead or cursed. it’s just not winning the current adoption race. hang in there. building real stuff on any chain is hard as ****.

• 
  Mahmoud Tantawy (@mtantawy) reported 23 minutes ago

  def not competitor but we're building a pipeline that would allow us to deploy when github is 100% down

• 
  Cristian Pena (@CristianPenaOK) reported 27 minutes ago

  Cursor just shipped Origin: agent-first *** hosting, S3-backed, 296K clones per hour, sub-400ms global sync. My weekend fix does the same job for most teams in 688KB and zero seconds. The gap between those two is the interesting part. A few months ago my coding agents kept stepping on each other in ***. I asked engineers I trust. Nobody else was seeing it. I spent a week assuming I was wrong before I solved it myself. Worktrees are the common parallel-agent move. The problem: worktrees share one .*** directory. Agents contend on refs, the index, and a single HEAD. They clobber each other. Fix: give each agent its own full clone via `*** clone --reference` against a local mirror. Shared object store, so objects don't copy. Isolated refs, index, HEAD, working tree. On the repo I was working in: 82MB mirror one-time, 688KB per clone, zero seconds. Main clone stays read-only via a guard hook. Agents coordinate via pull requests. Works on plain GitHub today. Zero new infrastructure. I solved the client side. Cursor Origin solves the server side. Different layers. Origin's value shows up when the *** host becomes the bottleneck: hundreds of agents hitting the same remote per hour. That's real at org scale. Not most individual developers' problem yet. We converged on the same conclusion from different directions. Where does plain *** stop being enough?