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:

• 
  CodeXSavage (@CodeXSavage6s) reported 11 minutes ago

  @rodrgds I still haven't run into any trouble with my GitHub

• 
  XLion (@XLionc) reported 12 minutes ago

  @FullStackNate @theo Terrible UX, not support GitHub Actions syntax

• 
  Ben Vargas (@benvargas) reported 18 minutes ago

  @AgofureDan @burkeholland @github Have a zen login and put a nominal amount in, 20 or 30 bucks of which I've used some... then when you have an active account like that you have an API Key from your dashboard and follow their standard documentation. Free models use don't reduce your wallet balance.

• 
  Lorenzo (@lorenzolfm) reported 18 minutes ago

  @PierreJoye @github What? Who the hell said anything about non-funded? Also, I want to run my CIs on my bare metal server. I do not wish that someone hosts this for free for me. Your failure to interpret what I said is hard to describe diplomatically :)

• 
  Utkarsh Kumar (@kumarukutkarsh) reported 19 minutes ago

  So, GitHub was down due to incomplete auto-scaling policy with retries worsening the outage.

• 
  Siddhant (@siddhant_borse) reported 19 minutes ago

  @HeyAliux GitHub copilot is down for your enterprise please contact the admin

• 
  dominioscripto.eth (@septiembre_eth) reported 20 minutes ago

  3/ What just happened. OpenRouter added “Sign in with Ethereum” to its login screen. Next to Google. Next to GitHub. Millions of developers can now authenticate with their .eth wallet on the world’s largest AI platform. No password. No email. Just your Ethereum identity. (3/5)

• 
  Unique (@uniquewithsauce) reported 20 minutes ago

  @kobixyzHQ @RialoHQ Do I need to sign in with my Github in the Agent Grand Prix homepage?

• 
  enp2s3 (@enp2s3) reported 23 minutes ago

  Free is not actually free. Here’s the Brave slop from my perspective. Popups everywhere, useless side features constantly jumping in your face like the browser is trying to force you to use them. When you first install Brave, if all you want is a better browser, you have to turn off dozens of settings and features first. Brave Ads, Brave News, Brave Rewards, yeah, earn rewards, because apparently we’re playing a ******* game here instead of looking for a browser, Brave AI Chat, Brave Talk… It genuinely starts feeling like one of the market's spam-filled products. Anyway, after turning all that crap off and getting through the initial setup chaos, you finally think you can relax. But you can’t. Brave looks good, the UI is solid, almost Opera-like hype, and it comes wrapped in all those big privacy-focused slogans. But after actually using it for a while, the whole thing starts turning into clickbait. A lot of Brave settings are disabled by default or set to ask every time, so on popular websites and across the modern web, instead of making your browser experience easier, it keeps costing you extra time. Still, you don’t quit Brave. You spend months fixing settings site by site, changing permissions, adjusting preferences, until you finally reach the comfortable browser experience you should’ve had from the beginning. Then not even 3 or 4 months later, another surprise. After a few browser version updates, basic usage starts falling apart with lag, freezing, stuttering and random slowdowns, gpu acceleration issues. At that point, the whole Brave experience basically becomes this: 1) Use Brave for a while. Spend a few months improving your settings and site preferences until you finally reach the normal browser experience you were supposed to have in the first place. 2) Just when everything finally starts working properly, one of Brave’s chronic problems randomly shows up. 3) Spend a few weeks researching it on github/reddit, trying fixes, taking backups and repeating the similar solutions until you eventually discover that people have been discussing the exact same problem for years and it is somehow still there. 4) Keep your hopes up, reset the browser, go back to Step 1 and start the whole thing over again. 5) After 1 YEAR, or X YEARS, finally find the courage to get rid of Brave and go back to Safari or Chrome and quietly weep. We basically have the browser version of the operating system irony now. The only difference is that we don’t hear about some brand-new operating system every year, while plenty of companies can build another browser, give it amazing PR and make huge promises about privacy, speed and simplicity. And as usual, we users are left quietly weeping.

• 
  MMM 🦠👽🫧 (@mdmadna) reported 27 minutes ago

  @sama Do you have a new product manager? Every update is terrible 1. Context menu for file uploading just drops down under the keyboard and basically inaccessible 2. GitHub plugin connection fails every time 3. Last update terrible UI update not fixed. It was perfect all black now💩

• 
  swathi.bhat (@swathidbhat) reported 30 minutes ago

  Day 5 of using @bot only on my phone. I had an idea which needed a new repo. But the bot couldn't spin it up. It guided me towards logging in GitHub via computer use within the app. I don't remember my login, man. Password manager please 🙏 I went off-app to create the repo myself.

• 
  Harshil Tomar (@Hartdrawss) reported 32 minutes ago

  How the hell do i make an AI Agent ? here is the full 2026 Sauce : scope - write the agent's single job in one sentence and pick the one metric / KPI that matters really - map the task as a flowchart first and try to break it down into as much of sub workflows as you can with bottleneck info - collect 10-20 real feedback of the department / team member doing this job today and try understanding the workflow with context foundation - start on a frontier model to prove the task is solvable at all before optimising anything - pick one framework and stay in it, openai agents sdk/ pydantic ai or langgraph all work - put the job, the constraints and the refusal rules in the system prompt and keep things onto github tools - give the agent a closed tool list declared in code with a typed schema per tool - talk and figure out monthly budget allocation internally and work backwards to the tools you can use for the same - wrap each external api in tenacity with exponential backoff and a deterministic fallback path tip: pass an idempotency key on every write so a retried call cannot duplicate the row knowledge and state - chunk your docs and retrieve from pgvector or turbopuffer ( can also try HydraDB ) instead of pasting context into the prompt - persist conversation state in postgres and rehydrate only what the current step needs - summarise older turns into a running memory record once the window starts filling control loop - cap max steps in the loop itself rather than trusting the prompt to stop - meter tokens and dollars per run in langfuse and alert on outliers - gate sends, charges and deletes behind an explicit human approval step - return "nothing found" on an empty tool result and stop the model from filling the gap - keep user text in its own message role away from your instructions evals - build the eval set from production transcripts your users actually sent - grade with an llm judge on correctness plus an assertion on every tool call that had to happen - run promptfoo or braintrust in CI and block the merge on a regression ship - launch behind a feature flag to ten users you can call on the phone - propagate one OpenTelemetry trace id from inbound request to final tool call - rate limit per user id with a token bucket in redis - queue work in temporal when a provider degrades so requests survive the outage - keep the kill switch flippable without a deploy after launch - try to observe and read full traces every morning for the first 1-2 weeks - move the easy intents down to a smaller model once the evals hold - give the agent a named owner and a dashboard showing runs, failures and cost it might sound complex right now but once you run through the motion of things; trust me, it will start to feel more and more easier we build in this exact order before any client agent goes live. most teams start at tools and find out from a user.

• 
  Praneethzz (@Praneeth_zz) reported 33 minutes ago

  I added Github to probe and it found 5 real problems that devs had. I stopped trying to scrape X because it was too risky and connected to Github using the official API instead. In the first run: -scanned 8 issues -Matched with 5 problem signals -extracted those 5 complaints -0 API errors I haven't really looked into the problems that devs have and post about in github, so this was definitely interesting to see.

• 
  MJY (@omninomsky) reported 34 minutes ago

  @Tech_girl How would it be able to determine whether the code was generated by AI or people? Realistically what would happen if GitHub would have to spend resources to determine whether the code was generated by a person (sounds expensive). It's very easy to defeat the purpose by being too sensitive (blocking human posts) too lax (allowing too much AI code through) or simply incorrect (blocking humans, allowing AI). It's very hard to get it exactly right. So just from the technical problem alone they probably won't do this. But let's say they did. How does that work out? Well, now people who rely on AI to do their work can no longer use GitHub. They'll probably want to keep using AI so they'll no longer use GitHub. I'm not yet seeing any upside. Less merges per day maybe? Relief for the system?

• 
  Google G-Ware Surface Technology (@GoogleWaveTech) reported 36 minutes ago

  @sama One honest distinction: each share creates a ledger commit inside StarQuest’s device-local wallet, not a new GitHub repository commit. Letting a public webpage commit directly to GitHub would require exposing credentials. Cross-device/server permanence still requires