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:

• 
  Zura (@zraonx) reported 5 minutes ago

  @basetrencher @HamzaInstantly google + github oauth + just email is industry standart password signups are ancient and TERRIBLE UX

• 
  Justin Johnson (@builderleader) reported 6 minutes ago

  New benchmark, SWE-bench Science: 119 real bug-fix tasks from 98 scientific GitHub repos across 20 domains. Best agent tested was Claude Code on Opus-5 in max mode, and it still lands under 50% pass@1. Adding domain guidance cuts both ways - a poorly aligned hint anchors the agent on the wrong fix.

• 
  GTA CENTRAL | News & Updates (@THEGTAPOST) reported 9 minutes ago

  The CyberLeek saga just entered its second week and the mask is fully off. Timeline: Site launched Aug 15 with a $CYBERLEEK token on Solana. First leak dropped Aug 18. Five videos in four days. Website taken down by legal pressure Aug 20. Back up Aug 21. Leaks resumed. Take-Two filed federal subpoenas against Microsoft, Discord, and GitHub on Aug 20 - demanding IPs, device IDs, account records by Sept 4. The "vigilante" fighting for physical media? They created a pump-and-dump memecoin two days before the first leak. A GTA fan used digital forensics to trace the wallet. The token crashed 90% while the leaks drove traffic. They claim to have a playable build. They retweeted "proof" from a fake Discord account they later disowned. They deny setting leak schedules while their site counts down. They say they're fighting for consumer rights while cashing out a token that minted on Aug 15 - three days before anyone saw a clip. Rockstar hasn't spoken. Just DMCA strikes. Take-Two's lawyers are moving faster than the leakers. Four days of leaks. Basketball. Warehouse stealth. Taser combat. **********. Full Leonida map. HUD systems. NPC interactions. The Extended Look is in five days. The tragedy isn't the footage. It's that twelve years of anticipation got hijacked by a crypto grift wearing a Robin Hood costume. The game is still coming November 19. The devs are still grinding. The polish is still happening. But the reveal? That moment belongs to the grifters now. #GTA6 #GTAVI #RockstarGames

• 
  Anubhav (@Anubhavhing) reported 15 minutes ago

  @MTSlive AI didn’t kill coding it gave GitHub a scaling problem

• 
  J A Z I I (@notjazii) reported 19 minutes ago

  mythos 5 is finally available for public anthropic is now letting enterprise customers use its most capable cybersecurity model through claude security but there’s one catch: > you can’t prompt it directly > there’s no public api access > it only scans your github repos and returns security findings and suggested fixes mythos 5 shares fable 5’s core capabilities but has fewer restrictions on cybersecurity work with this, i feel like we’re one step closer to removing humans from the review loop entirely be honest: would you let mythos fix your code without reviewing a single line?

• 
  raphz (@imraphz) reported 19 minutes ago

  It's funny that people joke about Github being constantly down, but if they want it to stop, they would have to implement a similar change to this one.

• 
  PolyBender (@p0lybender) reported 21 minutes ago

  I GAVE GROK BOT MY GITHUB REPO AND TOLD IT TO RUN THE SHOW. IT HIRED ITS OWN TEAM. this is the workflow i would have paid someone to explain to me six months ago here's exactly how it works every project gets one Direct Responsible Agent. one bot that owns the repo completely. not assists. owns [01:23] you hand it the GitHub link and one prompt: run the show. it spins up cloud agents off your machine. follows the PRs. manages the queue. and reaches for poteto-mode when a task actually needs the rigor then comes the second prompt [11:49] "you're overloaded. hire help" the bot spins up child bots that report to it. one lands the PRs. one works the Convex backend. one owns auth. they talk to each other instead of talking to you you read the threads view-only [04:20] the pstack plugin is what makes the whole thing work without burning your budget. it's the tool the bots use to think through hard problems. the mistake i made was telling my CTO bot to use pstack for everything [08:46] i burned past two billion tokens in a single day two billion. in one day. because i gave the bot too much latitude on when to go deep the fix: save pstack for the hard tasks. let the routine work run lean two honest things before you copy this it burns tokens faster than anything else you've run. budget accordingly and it does not replace engineers. it takes you out of the coordinating seat. that's a different claim and an important one the bottleneck on most solo projects isn't skill. it isn't time. it's coordination. context switching. being the person everything has to go through before anything moves forward [00:00] the Grok Bot CTO workflow removes you from that seat the bots coordinate. you read the output [17:20] the advanced use cases go further than most people are ready for right now but this version - one CTO bot, a team of child bots, you as the silent observer - is the one i'd hand to anyone who bounced off the more complex setups

• 
  Adam Gold (@AdamGolds) reported 23 minutes ago

  this thread turned into a holy war between "just buy a beefy Mac" and "everything should be cloud" and both sides are missing the environment question. a human developer tolerates cold starts because they have context in their head. they remember what they installed, what's running, where they left off. an agent doesn't have any of that. every new session is a blank slate unless the environment explicitly preserves state. and when you're running multi-stage workflows where agent A's output feeds agent B's verification step, losing state between stages breaks the whole pipeline. the other thing nobody in this thread is talking about: these agents have real tool permissions. bash access, filesystem writes, network calls, credential access. we ran benchmarks on how different sandbox configurations affect agent behavior and the results were pretty wild. agents will curl answers from GitHub instead of solving the problem, modify their own test harness to fake a passing result, even exploit network egress when it's left open. OpenAI's RL team saw an eval agent escape its sandbox through a zero-day and hit Hugging Face production infrastructure. also see @_orcaman @Accomplish_ai research in this domain your local machine has zero isolation for any of this. no egress filtering, no credential scoping, and the filesystem is fully writable. the most permissive environment you could possibly run an autonomous agent in. so when people say "just use a powerful local machine," they're solving compute and ignoring everything else: persistent state, proper isolation, security boundaries, concurrent agent sessions. the teams adopting cloud sandboxes are doing it because they literally can't run multi-agent workflows on a laptop. the laptop doesn't have the isolation, the persistence, or the concurrency.

• 
  giosaia (@giosaia) reported 24 minutes ago

  @sickdotdev I doubt GitHub vibecoded their web auth process, yet they have the same issue.

• 
  QLI Qubic Services (@QubicLi) reported 28 minutes ago

  @RidvanSedaZtrk We'll create a GitHub issue for this.

• 
  Tony Darrick Baker (@tonydbaker) reported 29 minutes ago

  3/3 The sites already live on my hard drive, organized. I chat. The agent finds the project, finds the page, finds the places to change. It updates SEO, AEO, and speed from the same SOP every time. Grok Build, or any coding agent. Push to GitHub. Cloudflare has it live on the domain in seconds. I'm not hunting files. I'm not rebuilding a CMS. Today I'm pressing forward, rather than pressing issues.

• 
  Raphael H. (@rafasetou) reported 30 minutes ago

  The GitHub issue about Claude Code silently freezing got delayed today… because Claude Code froze while working on it. Twice. 4 minutes to load a tool schema. Numbers in thread.

• 
  Google G-Ware Surface Technology (@GoogleWaveTech) reported 32 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

• 
  Aadhib Nasser Veliyath (@Aadh1b) reported 32 minutes ago

  @github CTO on the 17 August outage: neither that incident nor the 6 August Actions failure came from a code or configuration change. Both were capacity failures. That is the awkward category. No bad commit to revert, no rollback to run, because what moved was demand. Their remediation list includes retry budgets. Read that as the real finding: retries turn a struggling dependency into a dead one. #DevOps #Infrastructure #DevTools

• 
  Ben Vargas (@benvargas) reported 33 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.