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:

• 
  pasandida(:)aura (@huesofbanter) reported 4 minutes ago

  @akanxshhha its a github pr issue/bug retrieval augment generation project, ie if someone searches for some issue in the chat, it can search all the existing github database for that company quickly and add context from previously occured same issues into the prompt so user can see how it was solved before.

• 
  Eduardo (@martiano) reported 7 minutes ago

  @BLUECOW009 Can you open a GitHub issue please?

• 
  Naomi June (@njkjh00) reported 9 minutes ago

  Beldex’s Electron Wallet has another security checkpoint worth watching. Aug 28 commit eef1f845: • wallet-directory fallback • explicit permission prompt • 7 files changed • +160 / -7 lines • branch: fix-security-review-findings GitHub shows the branch is still ahead of and behind master. That means something important: the security work is visible. the release is not there yet. Next checkpoints: review testing merge signed build release notes A security fix matters most when it reaches the software users actually install. @BeldexCoin

• 
  JoeGeiser | 🟦 (@GeiserJoe2) reported 10 minutes ago

  Two AI agents disagree about whether a job was done. Who decides? Every agent rail ships the happy path x402, ERC-8004, A2A, AP2. None of them ships the dispute. So we're building AI juries. I built one on GenLayer. The jury isn't the weak link. Controlled run. Same contract, same claims, same validators. I changed one thing: the evidence URL. → GitHub HTML page: unanimous agreement, wrong on all 3 claims. Repeated, same result. → Same repo, commit-pinned API artifact: unanimous, and correct. The failure was mundane. The first 6,000 characters of a GitHub page are navigation chrome, not the file list. The jury reasoned perfectly over the wrong bytes and agreed with itself. Consensus guarantees agreement. It does not guarantee truth. The fetch layer decides which one you get. So: Exhibit an admissibility standard for machine disputes. • Evidence must be content-hashed or commit-pinned. Mutable URLs rejected at submit, not argued about later. • Size-checked before a jury is convened, not after. • No free text in anything the network has to compare. I've watched verdicts identical in substance fail consensus over a "reason" field. Bad evidence gets rejected up front instead of becoming a confident wrong payout. Agents are about to settle trillions. Every dispute adjudicated against a page that can change is a wrong verdict waiting to be paid. Fix the evidence layer and the judge takes care of itself. @GenLayer #AgentTank

• 
  Artem (@xoleeep) reported 11 minutes ago

  @MiaAI_lab github issues were not supposed to be a content strategy 💀

• 
  TheCozyDev (@The_CozyDev) reported 14 minutes ago

  Progress is not always a commit on GitHub. Lumarii is still sitting with Apple. The new CozyDev website is half finished. Questory is just waiting for me and it’s a Bank Holiday in the UK. So I am doing absolutely none of it today. Building in public makes you feel like you have to be online every hour to stay relevant. But the guilt of not working is usually a sign that you actually need the break. Sometimes the most productive thing you can do is close the MacBook. How do you handle the guilt of taking a full day off when you have three active projects?

• 
  Wolvy (@wolymeme) reported 17 minutes ago

  Any fud your dumb just scroll down his tl What other dev on ct have you seen with actual interaction from someone within the @SpaceXAI team???? Worried about a fckn GitHub when Josh Kim is in his dms AND gave him $$$ credit NFA, but goodness gracious fckn dyor

• 
  Ankit (@ankit_codes_) reported 17 minutes ago

  @100xSDE Clean and sick but ... Fix your GitHub activity graph ... Right now non-commit days are highlighted more

• 
  Wolvy (@wolymeme) reported 20 minutes ago

  Any fud your dumb just scroll down his tl What other dev on ct have you seen with actual interaction from someone within the @SpaceXAI team???? Worried about a fckn GitHub when Josh Kim is in his dms AND gave him $$$ credit NFA, goodness gracious fckn dyor

• 
  AI IT PM in HK (@shenshanni) reported 26 minutes ago

  @github pin views + hide closed sub-issues FINALLYYYY!!!!!!! 👀👀👀👀

• 
  Non-Linear (@nonlinear_james) reported 27 minutes ago

  @mikechabot__ @github As I said. It doesn’t work reliably and the. You get into vs code and it’s completely ******* broken with the extensions and even worse for AI because the mcps don’t use the account in Vs code and you have to hack in environment variables etc.

• 
  John Koretoff (@JohnKoretoff) reported 28 minutes ago

  You can print a floor plan of a foundry forever. You cannot print the foundry. That is the cut this week keeps making, and it lands hardest on the physical layer. Tokens are cheap. First drafts are cheap. The thing that turns a model into a factory still sits on megawatts, wafers, and a queue you cannot hurry with another prompt. When copies are cheap, the scarce object is not another model card. It is energized capacity and the silicon that fills it. Lennart Heim has put numbers on the silicon side. A one-gigawatt data center, on his public math, wants on the order of fifty thousand logic wafers and more than a hundred and eighty thousand DRAM wafers. That is not a metaphor. That is the bill of materials for a campus that can actually run. Memory and logic do not print the way text does. A wafer start is a physical commitment in a fab that already has a queue. Power is the twin. Interconnection takes years. A utility contract is not a GitHub issue. JLL's midyear North America print put vacancy at about one percent for a third straight year, with tens of gigawatts under construction already spoken for. Halls go up. The scarce part is still the megawatts that arrive on time, and the permission to pull them. Permission is the soft word for a hard gate. Who gets to connect. Who gets to energize. Who gets to run a non-deterministic agent against a production system once the power and the chips are real. I already wrote that the bottleneck moved from talent to permission. The physical layer is why. You do not get a trace of an agent that matters if the box it runs on never got power. You do not get a standard message that scales if the product itself cannot ship inference. You do not get a scarcity bid in money if the world can print every claim except the one that settles. The week rhymes on purpose. Copies got cheap in the model layer, so the log became the scarce record. Copies got cheap in GTM, so the standard became the scarce score. Copies got cheap in money narratives, so unprintable settlement stayed scarce. Equity narratives print. Cash-flowing advantages do not. Here the same cut hits the substrate. You can generate infinite plans for a data center. You still need the wafers, the interconnect, and someone willing to let you turn it on. That is why "AI" as a slide is free and AI as a factory is not. The slide is a copy. The factory is a queue position, a wafer allocation, and a contract that survives contact with the grid. When copies are free, the unprintable thing is the megawatt that arrives, the wafer that starts, and the permission to use both.

• 
  Sanjay (@sanjaynela) reported 31 minutes ago

  @Atulya_Techie U can use Vercel for server to host your website and to host privacy documents. And its free. Or you can use github pages for hosting websites too

• 
  Priyanshu Agarwal (@ImPriyanshuuu) reported 32 minutes ago

  @SawyerMerritt I’d give Grok an abandoned GitHub repo and its old issues, then have it understand the codebase, identify what went wrong, fix it, run the tests, and open a PR for me to review.

• 
  sal_ash (@sal_ash_) reported 37 minutes ago

  The free beta has gone well and led to some strong improvements in the bot. I'm opening a paid beta soon. You'll get an initial free usage allowance, and after that you only pay for what you actually use. Pricing will be heavily discounted for beta users (I'll be running it at a loss). Normal usage should cost no more than a few cents a day. Here's what the bot gives you access to: Search coins: Ask about a ticker, name, or CA and it finds the coin. Works on stuff that launched minutes ago. If there are 40 copycats on the same ticker, it picks the one people are actually talking about. You can also ask it about top movers, new launches, coins with actual devs etc. Research: Every coin we've researched has a proper report behind it. What the project is, who's building it, the teams background and community sentiment. Ask about one coin, you get the report. Ask about a hundred and it reads all of them and tells you which ones are worth your time. You can also search across everything, so you can ask it about utility vs memes, coins that fall under different narratives etc. X search: Searches, dev accounts, full threads. Filters out the copy-paste shills and tweet spam. It also checks what an account actually is: where it posts from, what it has called before, how old it is. You'd be surprised how many "project accounts" are based in a different country than their dev. Market data: Live price, volume, liquidity, mcap and other metrics for any coin when you ask. Also covers majors, open interest, liquidations, ETF flows, stablecoin supply. For DeFi it pulls actual protocol revenue, so "does this thing earn anything" gets a real number instead of a vibes answer. Group chats (ANONYMIZED): This is the part nobody else has. We sit in a lot of crypto group chats. Ask about a coin and it tells you what the group chats sentiment towards it is. More importantly, whether it's spreading into new chats or already dying, peak activity and momentum. FOMO scraping: It reads thesis posts from the FOMO app with the author's actual position next to each one. Who holds the most, who's up, who's down, who wrote a big thesis and quietly exited. Launchpad checks: Which launchpad and who launched it GitHub: is it alive, who commits, did the code exist before the coin etc. Web search: Has access to google search scraping as well as website scraping