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:

• 
  Wizard 🏴󠁧󠁢󠁷󠁬󠁳󠁿 (@shiddandfardd) reported 3 minutes ago

  @DIT545songs And if they are providing an exe, and github UI is the problem, then they should put a link to the download at the top of the readme on the front page. Solves the whole problem and makes it easy for everyone.

• 
  Kingmaxx22🕹️🔖 (@22Kingmaxx) reported 4 minutes ago

  @Double2MC @sophiiess_ yes this is the real issue making a website that links to a major release on github isn't that complicated

• 
  ⚡Tech Momentum⚡ (@TechMomentum_) reported 4 minutes ago

  Anthropic quietly gave Claude Code a real browser. Not a screenshot tool, a tabbed pane inside the app that the agent reads, clicks, and navigates on its own. Ctrl+Shift+B (Cmd+Shift+B on Mac) opens it. Ask Claude Code to check a library's docs, click through your own signup flow, or read a GitHub issue thread, and it just does it. No copy-pasting URLs into chat. Two guardrails keep it from going rogue: a classifier reviews every write action, and there's a domain allowlist outside Auto and Bypass modes. OpenAI shut its own browser tool down earlier this year. Anthropic went the other direction. If your agent still can't open a webpage on its own, that's not a limitation anymore. That's a setup problem.

• 
  WorktreeWise (@worktreewise) reported 6 minutes ago

  *** tip: Create worktrees outside your main source directory to avoid accidental commits or lint scanning issues. Example: `../worktrees/my-feature` #*** #GitHub #DevTools

• 
  Shai (Deshe) Wyborski (@DesheShai) reported 8 minutes ago

  Thanks for the new "paper" David, I could do with a good laugh. But I was talking about math, not about writing a bunch of mathy-looking stuff and posting it on your GitHub for your little fans to gush over. Full disclosure: I didn't read the whole thing. I didn't have to. I only got up to page 2 before running into a blatantly fatal error, and in page 4 I found another one, so I didn't see a reason to push forward. Maybe that's why you had to present it in workshop instead of trying to get it published. Anyway, let's get to it. In page 2, you said “Compensating for this decline […] requires the attacker to add a supplementary transaction containing real fees.” That's false. It does not follow from requirements 1-5. This assumptions hide the fact that your "analysis" implicitly assumes that attackers always try to preserve the same-production work. That is, your "proof" only prohibits one attack vector, and disregards most realistic scenarios such as: - Routers that insert Sybils without producing the block - Attackers whose block already contains enough work Explicit example: without Sybiling at n=H=2, the atataacker routing share is u(2,2) = 1/3. After adding one identity, we have u(2,3) + u(3,3) = 2/7 + 1/7 > 1/3 so the attacker controls hops 2-3. That is, whenever this transcation is included (which has unchanged positive probability. E.g. if the eventual block producer already satisfies the work threshold from other transactions, inserting this additional hop need not change the transaction’s probability of inclusion), Sybilling strictly increases routing payout for free. That's not the only gap I found. You derived propagation asa preferable only when x > y/7, but then kinda just assumed this means y>7x is "impossible". More egregiously, you apply the inclusion-exclusion principle while switching between the distributions x and y. These distributions might have the same sample space, but a different probability measure induced by different strategy. Hence, using inclusion-exclusion to establish a relationship between them is a rookie mistake which completely invalidates your proof strategy. This invalidates your entire proof strategy. Applying the exclusion-inclusion principle while using a different probability function for each event is a rookie mistake. Your paper might prove that "compensated Sybil" is a bad strategy (which is not exactly a surprise. Why would anyone want to do that?). But it's a far cry from "disproving Babaioff" and is exactly what happens when cryptobros think they are mathematicians. They do a lousy job all around, and post it to their GitHubs so that their little following of people who think they are going to make them rich will laud how smart they sound to the uninitiated. Either way, having found an actual deep flaw in your "work" (thanks for bringing it up!), I'll mail my concerns to IWGTEA, which might save them some embarrassment. But again, thanks for the laugh. I needed it.

• 
  Gaston Levai (@gastonelevai) reported 9 minutes ago

  @GRAZZLEYNFT @PumpfunEco After also the community address the issue with the OG dev he finally redirected the fees to the GitHub of @PrJulienMartin .

• 
  s1r1us (@S1r1u5_) reported 15 minutes ago

  let me present a case of ai-induced overconfidence. btw, if your company is doing this, call it out. it is a dumb assumption, and let me tell you why. one of your security leads or engineers starts using claude code or codex, or builds a cool wrapper around them. it finds some bugs, of course. they run it repeatedly, burn **** ton of tokens, fix everything it finds, and eventually the scanner goes quiet. the next thing that happens is they decide the bug bounty program is a waste of money. after all, how could some random hunter find anything their sophisticated pipeline missed? what could an external auditing firm possibly find that their agents couldn't? so they cut the program. but if you are truly confident that your pipeline has exhausted the vulnerability space, shouldn't you be willing to increase the bounty? wouldn't that be the best way to show your c-level bros that the ai spend actually worked and wasn't wasted on some wrapper you vibecoded in a week? or..... maybe you don't want to incentivize people to test that assumption. because how would you explain to your boss that, after burning through all those tokens, some bug bounty hunter still found a nasty vulnerability? your cool wrapper finding bugs does not prove that no bugs remain. it only proves that the wrapper can find subset of vulnerabilities and researcher with taste will always find a way in. the entire point of a bug bounty program is to incentivize skilled researchers to find the bugs your internal tools, agents, and engineers missed. think about it, why do openai and anthropic still run bug bounty programs that pay $100k for critical vulnerabilities? they have effectively unlimited tokens and run all kinds of crazy agent loops and automated security workflows. yet they still sometimes pay that $100k, because software is vast and complex, and someone with taste will always find a bug your pipeline missed. remember, researcher with taste + ai(this is where your threat actors are) > good wrapper > plain claude code/codex if your public program is drowning in slop, move it to private. invite strong researchers and pay them vip prices like github is doing. don't have this ai competence illusion which is pretty common all around.

• 
  Cool Guy™ 🌈✞🌛 (@ChurchOfMoons) reported 15 minutes ago

  It's kind of nuts how GitHub people just...don't know how to communicate with people. I worked with it for years, I don't have a problem with it, but people keep putting simple executables on there and getting surprised when people just want to one-click download something.

• 
  s1r1us (@S1r1u5_) reported 15 minutes ago

  let me present a case of ai-induced overconfidence. btw, if your company is doing this, call it out. it is a dumb assumption, and let me tell you why. one of your security leads or engineers starts using claude code or codex, or builds a cool wrapper around them. it finds some bugs, of course. they run it repeatedly, burn **** ton of tokens, fix everything it finds, and eventually the scanner goes quiet. the next thing that happens is they decide the bug bounty program is a waste of money. after all, how could some random hunter find anything their sophisticated pipeline missed? what could an external auditing firm possibly find that their agents couldn't? so they cut the program. but if you are truly confident that your pipeline has exhausted the vulnerability space, shouldn't you be willing to increase the bounty? wouldn't that be the best way to show your c-level bros that the ai spend actually worked and wasn't wasted on some wrapper you vibecoded in a week? or..... maybe you don't want to incentivize people to test that assumption. because how would you explain to your boss that, after burning through all those tokens, some bug bounty hunter still found a nasty vulnerability? your cool wrapper finding bugs does not prove that no bugs remain. it only proves that the wrapper can find subset of vulnerabilities and researcher with taste will always find a way in. the entire point of a bug bounty program is to incentivize skilled researchers to find the bugs your internal tools, agents, and engineers missed. think about it, why do openai and anthropic still run bug bounty programs that pay $100k for critical vulnerabilities? they have effectively unlimited tokens and run all kinds of crazy agent loops and automated security workflows. yet i bet they will pay that $100k, because software is vast and complex, and someone with taste will always find a bug your pipeline missed. remember, researcher with taste + ai(this is where your threat actors are) > good wrapper > plain claude code/codex if your public program is drowning in slop, move it to private. invite strong researchers and pay them vip prices like github is doing. don't have this ai competence illusion which is pretty common all around.

• 
  Mr Momoh - Brother Ridgeback 🦁 (@tonylab_net) reported 16 minutes ago

  Reviewing a GitHub PR shouldn't mean copy-pasting the diff into a chat window. So I built this: hit ⌘⇧K on any PR, file, or issue and Kimi K3 reviews, explains, or summarises it — right there, streamed into a side panel. Free. Open source. Bring your own key.

• 
  Ask GPTs (@askgpts) reported 18 minutes ago

  Lightpanda built a headless browser from scratch, no Chromium, no forking anything. Just javascript execution and web APIs, stripped down to what agents actually need. Chrome costs you 207MB of RAM and 25.2s to hit 100 pages. Lightpanda does it in 24MB and 2.3s. Here's what makes headless automation hard normally. Every scraper and agent framework has been dragging Chrome's entire rendering engine around just to read some HTML and click some buttons. Blink was built for humans watching pixels, not for bots reading data. → Puppeteer & Playwright: drop-in compatible through CDP → Agent mode: plain english instructions, it navigates and extracts → PandaScript: /save turns any agent session into replayable JS → MCP: native support out of the box Here's the wildest part. You use the LLM exactly once to figure out the flow. Then you export it as a script and run it forever with zero model calls. The agent trains itself out of a job. Here's the cost comparison that makes this worth caring about. 11x faster execution, 9x lighter memory, and $0 in ongoing inference cost per automated workflow after the first run. Written in Zig. AGPL-3.0. 12k GitHub stars. 100% Open Source. GitHub link in the comments

• 
  Jb (@Jb21mm) reported 18 minutes ago

  There is a serious issue with offshore devs using GitHub copilot to code for them. And then they are too dumb to know their code has issues. Because offshore brethren also QA with the same GitHub ai...

• 
  Laughless (@LaughlessVR) reported 19 minutes ago

  @CryptoCyberia This issue isn't really with github, it is with developers using github as a distribution method for end users who are unfamiliar with the UI, who would have no reason to learn to navigate said UI if the developers used a proper dedicated distribution method.

• 
  Wim (@Wiim1986) reported 19 minutes ago

  @CryptoCyberia Tbh, I always have trouble finding the installation package as well in github. Like, it used to be easier, even when you had to dodge fake download buttons on crummy sites.

• 
  nd-m (@Andreas94024677) reported 24 minutes ago

  @thsottiaux automatic review on GitHub is to slow.