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:

• 
  ✚ Oz ✚ (@oz_a13banger) reported 7 minutes ago

  @jediahkatz Im super interested in this. Im especially hopeful they have something better than GitHub Issues for tracking things. We use Issues heavily but have always been unhappy with it.

• 
  AcceptÐoge (@DogeAccept) reported 9 minutes ago

  @BuildrJ @DogeOS to start: 2025/09/15, you blocked me for asking you about the zk L1 proposal that Jordan submitted to the $doge project's repository on Github. Prior, Jordan also blocked me for asking about his proposal that's been there for a year. Is there an issue in talking about L1?

• 
  Luke The Dev (@iamlukethedev) reported 10 minutes ago

  GitHub is in trouble. Cursor already owns a huge part of the coding workflow. Now they’re building the repository too. The closer AI gets to the code, the less room there is for everyone else.

• 
  Ferbin (@Ferbin08) reported 12 minutes ago

  @swyx @TomasReimers @cursor_ai You could build better UX, better search, whatever. The product isn't the bottleneck. The real issue: every developer's collaborators are already on GitHub. That's the moat no competitor can beat.

• 
  Cristian Pena (@CristianPenaOK) reported 13 minutes ago

  Cursor just shipped Origin: agent-first *** hosting, S3-backed, 296K clones per hour, sub-400ms global sync. My weekend fix does the same job for most teams in 688KB and zero seconds. The gap between those two is the interesting part. A few months ago my coding agents kept stepping on each other in ***. I asked engineers I trust. Nobody else was seeing it. I spent a week assuming I was wrong before I solved it myself. Worktrees are the common parallel-agent move. The problem: worktrees share one .*** directory. Agents contend on refs, the index, and a single HEAD. They clobber each other. Fix: give each agent its own full clone via `*** clone --reference` against a local mirror. Shared object store, so objects don't copy. Isolated refs, index, HEAD, working tree. On the repo I was working in: 82MB mirror one-time, 688KB per clone, zero seconds. Main clone stays read-only via a guard hook. Agents coordinate via pull requests. Works on plain GitHub today. Zero new infrastructure. I solved the client side. Cursor Origin solves the server side. Different layers. Origin's value shows up when the *** host becomes the bottleneck: hundreds of agents hitting the same remote per hour. That's real at org scale. Not most individual developers' problem yet. We converged on the same conclusion from different directions. Where does plain *** stop being enough?

• 
  saaslane (@saaslanee) reported 14 minutes ago

  @code_kartik where are u people when github is down and thousands of people complaining? lol no reason for anyone to leave?

• 
  John Crickett (@johncrickett) reported 16 minutes ago

  @CFDevelop I've tried many models and agents. Claude Code is the flakiest agent. Anthropic's model is the least reliable (sometimes down). I get good results from Codex, Kiro (multiple models), Github Copilot (multiple models) and OpenCode (multiple models including open source ones). Anthropic is the buggiest.

• 
  Hugo Ruiz (@HugoRz_) reported 17 minutes ago

  GitHub was built for humans. Origin is being built for agents. Different problem, different tool. Makes sense.

• 
  TokenFires (@TokenFires) reported 18 minutes ago

  Tired of Claude not working well? Me too. So I figured out how Anthropic has trained the model to expect to work. You work with Claude now, Claude does not work with you. Here are the keys to success with Opus and Sonnet: 1. Provide a strict set of agent instructions: - start with Karpathy’s rules - add run up and summary removals - add refusal for questions it can find the answers to - tune for preferences - enforce verification not assumptions - enforce responsibility (model performance will be discussed in retrospectives) - keep it SIMPLE though (aka: limit token burn and confusion for the LLM) - be specific about *** ops 2. Follow this workflow: [opus] research (docs/web = define source of truth) -> plan (intent and what success looks like) -> design -> task decomposition (target sonnet)-> create failing tests -> [sonnet] construction -> bug fix until tests green -> [opus] review against plan/design + test validation -> cover deploy/rollback. Then it works fine. Beats the 30 day rolling memory window Claude ships with. And/or, add a real memory system to Claude. Raw sessions + prompting went away with 4.5. Anthropic did not express this as strongly as they *could* have. But the 2026 versions expect a certain workflow now. If you work in it, it’s successful. If you skip anything or try to vibe your way to the end, it’s less likely to result in quality code. And your session will churn with flip flop changes and miscellaneous bug fixes. Claude *NEEDS* a library of good representative information to draw from through the whole process. Don’t skip doc building and providing web links with explanations (look here for this, read this for that). Try to shortcut this and the Claude models don’t “work”. Even better, build agents (or find built definitions on GitHub) that do these and create a skill walking through the whole process. I promise the result is better after the pre-work is done. I’m paid to do this and I ship AI code without the hype and vibes in my day job. Every week. Every day. Do people on X do this though? Is this a largely unknown thing outside of the software engineering field? Oh. And add hooks for delete and drop commands. And never connect AI to production. I feel like I shouldn’t have to say these things. But I know we’re only human.

• 
  outloop (@0utloop) reported 21 minutes ago

  I had a GitHub repo with my ex. I still create issues, but there's no response anymore.

• 
  Jason (@followjason) reported 22 minutes ago

  🔗 + TLDR Zodl 3.6.0 adds support for viewing wallet balances and payments in any of 23 fiat currencies instead of only USD, with automatic updates that respect Tor privacy settings. Server selection now offers simple Automatic (best available) or Manual options for reliable connectivity, while new wallets sync directly from the current chain tip to skip outdated checkpoint scans and speed up setup. Release incorporates community input including a GitHub pull request from tippenein for currency features and collaboration with Valar Group; update available via App Store, Google Play, and new F-Droid repository.

• 
  AIToolsDailyIN (@AIToolsDailyIN) reported 24 minutes ago

  @github The "AI is dangerous for security" camp has real data. Cybersecurity firm Snyk tested multiple AI coding assistants in 2023 and found consistent issues: hardcoded credentials, injection vulnerabilities, weak crypto defaults.

• 
  Carlos Rodríguez (@carlosrr004) reported 24 minutes ago

  Just when GitHub was having huge performance issues...

• 
  AI Mastery Guide (@aiseomastery) reported 24 minutes ago

  CLAUDE'S ENTIRE SYSTEM PROMPT WAS LEAKED AND ANTHROPIC CANNOT TAKE IT DOWN 27,000 tokens of hidden AI instructions are now public on GitHub.

• 
  shmidt (@shmidtqq) reported 25 minutes ago

  Creator of Claude Code: "Right now you still need to know how to code. In a year or two, it won't matter. I haven't edited a single line by hand since November." In a 90-minute podcast, Boris Cherny breaks down the exact setup behind the tool now writing 4% of every public commit on GitHub. More value than a $500 vibe-coding course. Save this. In a year we'll know if he was right.