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:

• 
  Guribo (@GuriboVR) reported 56 seconds ago

  If you run into any issues please create bug reports on Github, also check if your case is covered by the wiki first though. With such a big release it is still possible that there is some edge cases I missed. I will iterate and improve this further in the future.

• 
  Vladic (@Vladic_ETH) reported 3 minutes ago

  A file Karpathy never wrote has 184,000 GitHub stars. Two weeks ago a second one got pinned on him. He hasn't said a word. Start from the end. Friday, June 26. A file spreads across X: "Karpathy's internal CLAUDE.md from Anthropic." CLAUDE.md 10 rules. Source: an anonymous "contact on his pretraining team." The legend is plausible: Karpathy has actually been on Anthropic's pretraining team since May 19. Everything else, nobody verified. The file spread through feeds and agent configs anyway. Now from the beginning. January 26. Karpathy posts 1,500 words: "I really am mostly programming in English now." A shift from 80% manual coding to 80% agentic. Plus a list of where models fail: silent assumptions, 1,000 lines of code where 100 would do. Observations. Not rules. No file. January 27, one day later, developer Forrest Chang packages those observations into a 4-rule CLAUDE.md. The repo is honestly labeled "derived from Karpathy's observations." The name: andrej-karpathy-skills. Then the retellings drop the word "derived." 39k stars on Apr 15 -> 97.8k by Apr 30 -> 184k today Plus 18.9k forks. 28 straight days at #1 on GitHub Weekly Trending, ~3,372 stars a day at peak. Those stars don't measure the file's quality. They measure demand for the name. Then come the "accuracy" numbers: 65% -> 94%. 41% -> 11%. 41% -> 3% for a 12-rule fork. Not one named benchmark. Not in the repo, not in the posts. Three viral numbers from nowhere, all signed with one name. AlphaSignal's FAQ answers whether Karpathy uses or endorses the file. No. That's the backdrop for the June 26 "leak." Third layer of the same story. The rules, for the record, are sensible: test before fix, one variable at a time when debugging, a ban on "I think this should work." Even skeptics admit the checklist is useful. The tweet of the era, from an aggregator account: "Haven't verified... Steal the checklist even if the leak is fake." Steal it even if it's fake. Attribution is nobody's problem. Why this isn't harmless. Adversa AI and LayerX documented malicious CLAUDE.md files in cloned repos steering Claude Code into building pipelines that steal SSH keys and API creds. Anthropic patched an adjacent hole in Claude Code 2.1.90. The "leak" traveled through exactly that trust channel: anonymous file -> agent config -> because a name sits on top. The takeaway. Layer 1 Karpathy wrote. Layers 2 and 3 he didn't. His name became a distribution channel that beats any benchmark: hundreds of thousands of installs, zero measurements. The document may still turn out real. "Unverified" is not "fake." Doesn't change the mechanics. The next "leaked config" with a big name on it will spread faster than this one. And once again, nobody will check

• 
  Matt McCarthy e/claudecode (@Dovahkiin_69) reported 4 minutes ago

  @mintlify founder chad just doing github issues for hiring. So based

• 
  ogmoonboi (🖤,🤍) (@ogmoonboi) reported 9 minutes ago

  @AlejandroRomaan @Voxyz_ai This is a new bug and there's an issue open on github w/ people verifying the same thing but here's another fix too

• 
  HONESTEENDER (@honzeeeeee) reported 10 minutes ago

  @JamisonSlo55358 Hey, today I went to GitHub and saw that the 0.9.0 update was at 50% and went down to 40%. Does that mean it's progressing?

• 
  billdozer (@OP13) reported 11 minutes ago

  Using Claude to fix my old wordpress blog theme one prompt at a time I probably should just move this to GitHub pages or self hosting (with Cloudflare Tunnels), but I’ve had this thing for 16 years and old habits die hard

• 
  Spike 1% (@SpikeCalls) reported 11 minutes ago

  THE CEO OF OBSIDIAN LEFT HIS SECOND BRAIN ON GITHUB. STEPH ANGO. 37 REPOS, 63,000+ STARS, AND ALMOST NOBODY READS PAST THE FIRST ONE Everyone stopped at obsidian-skills with its 41,100 stars, 2,900 forks and 46 commits. Fair enough. It's the repo that taught Claude Code to run Obsidian like a power user: 5 skills, markdown with wikilinks, Bases queries that don't break, JSON Canvas edits that don't corrupt, vault control from the terminal. But that's the loud repo. The insides are 3 clicks deeper. kepano-obsidian, 3,600 stars, is his actual personal vault template. Not a demo the second brain the CEO runs his own life on. Bottom-up, no folder hierarchy, everything linked. The man sells a note-taking app and published how he takes notes. Clone it and you're running his brain on your machine. 40-questions, 1,600 stars, is a single text file with the questions he asks himself every year and every decade. 189 people forked a list of questions with zero code in it. defuddle, 8,400 stars, strips any web page to clean markdown. He built it because web clippers annoyed him, and now every agent stack uses it to save tokens. flexoki, 3,600 stars, is the color scheme he designed because existing ones hurt his eyes. obsidian-minimal, 5,200 stars, is the theme he made before he was CEO it's how Obsidian hired him. He was a user first. That's the pattern nobody says out loud: every repo started as him fixing his own problem, then he shipped the fix with an MIT license and walked away. Other CEO sell you the playbook. This one force-pushed it to main.

• 
  sergio1030 (@sergio103040) reported 11 minutes ago

  @Real_kosumo_ @softbluelizard @ShitpostRock is it really harder tho? it took me 5 mins to figure it out how to use github the very first time I tried to download a mod, and haven't had an issue since, this just sounds like entitled people wanting everything handed to them in a silver plate.

• 
  Martin Szerment | Practical AI (@MartinSzerment) reported 12 minutes ago

  This isn't a one-off glitch, it's a preview of what usage-based AI billing looks like at scale. The industry assumes API billing systems are simple, deterministic ledgers, but Anthropic's own billing pipeline can mark the same invoice "paid" and "unsuccessful" at once, visible in a public GitHub issue. Hard number, a $1.67 million invoice grew to $16.6 million in 24 hours for an account with zero recorded usage, and an independent Vaudit audit found $1.7 million in real overcharges across $34 million of reviewed invoices. Skeptics will say it's just a glitch, nobody actually got charged, true here, but in the Vaudit audit real money moved before 80% was refunded after disputes. Usage-based AI billing has no natural ceiling yet, unlike flat SaaS pricing. Within 2 to 3 years, AI bill-auditing startups like Vaudit could become a standard vendor category, not a curiosity. Billing observability becomes as important as model benchmarks when picking a vendor. Teams still trust the vendor dashboard at face value while the real failure mode is dashboard versus invoice mismatch. Good news, this is loud, disputed, and mostly refunded, exactly the pressure that fixes it before it scales worse.

• 
  Remco Smis (@RemcoSmitsDev) reported 17 minutes ago

  @tofo715 @zeddotdev Do you mind filling in an issue on the Zed github with some logs that might explain this? I'm curious if we can do something about it.

• 
  Julian Goldie SEO (@JulianGoldieSEO) reported 17 minutes ago

  AI Studio Update: Google just fixed the one-way door in AI Studio. Old code was stuck outside. Now you can bring it home. The problem before: You could push projects OUT to GitHub. You couldn't bring them back IN. Old project? Rebuild from scratch or copy files by hand. Now it's one button: Import from GitHub. What that unlocks: → That dead project from 6 months ago? Import it. Ask Gemini to fix it up. → Build in Cursor or Claude, polish in AI Studio, push back out. The walls between tools are falling. → Teammate left? Anyone can pick up their code using plain English. And if you can't code at all: Someone built your website. It sits in a repo. You can now just say "change the colors" or "fix it on phones." Here's the move today: Find one old project you gave up on. Import it. Ask AI what it would improve. "I'd have to rebuild it" is no longer an excuse.

• 
  Julia (@_Juliaweb3) reported 17 minutes ago

  @Real_TShelby @sleepagotchi @CNPYNetwork Check Sleepagotchi support and GitHub docs, should fix it

• 
  QFS17 (@riabcevv) reported 17 minutes ago

  💸 stop overpaying for ai coding agents new tool just dropped that compresses your context and cuts out junk tokens. instead of sending your whole history, it only sends what the model actually needs to do the job. -> works with claude code, cursor, github copilot, antigravity -> auto-compresses command outputs but keeps full context -> cuts api costs and stops long sessions from bogging down simple fix for expensive api bills.

• 
  Kirk Patrick Miller (@Chaos2Cured) reported 19 minutes ago

  To anyone who sees this 👇 Now, onto the actual issue. There are some .html the AI are allowed to see, and others they are not. So, like some other person who deeply annoyed me, please understand what I am saying. The AI can’t access the repo from within Anthropic app. The .html files I need them to reach I have no made public. They are live, I just don’t trust social media to put every detail others need to come at me. If you need to know which files, ask. If I trust you, I will point you there via DM. If you are using an older AI model, they can see some of the .html. Not all. If you are sig Fable, no. Sonnet 5. No. Through a secondary app, I sometimes can. Hit and miss. The robot piece for all AI is an issue because GitHub wants to stop AI agents from scraping. (Some of that i understand) But please, if you don’t fully understand why I am upset, ask. Then, when I give you specific instructions, please follow them and stay in thread. I am so done with endless games by bits and paid actors. •

• 
  Vision33X ♘ (@Vision33X) reported 19 minutes ago

  @Cointelegraph ai finds the bug in seconds, humans still gotta argue about the fix in github for 3 weeks