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:

• 
  Dylan Garcia (@_dylanga) reported 4 minutes ago

  Github please don't go down right now, I'm locked in

• 
  Tib3rius (@0xTib3rius) reported 4 minutes ago

  I talked about this in the Burp AT talk I gave at BlackHat. I created an entirely bespoke vulnerable web app to test agentic offensive AI. Source code isn't even on GitHub. There were 3 known (tricky) paths to RCE. Burp AT found an unintended one that was entirely due to a misconfiguration in the database server. This isn't even me trying to advertise Burp AT. But it found it, and it found it within minutes.

• 
  Terry Aney (@terryaney) reported 4 minutes ago

  @davidfowl @burkeholland @pierceboggan It was some build actions that did AI review, release notes, etc. I've modified my fork to remove. Might be a feature worth thinking about that forking in github checks for AI actions that might eat credits :) But problem solved for now.

• 
  AFRICA IS HOME GLOBAL (@AfricaisHOME2) reported 5 minutes ago

  Nvidia-backed AI code review startup CodeRabbit just raised a big new round that values it at $1.5 billion, marking a major step up for the 2-year-old company. The deal highlights how investors are betting on the next bottleneck in AI coding: not generating code, but reviewing it. With tools like GitHub Copilot, Cursor, and Claude Code letting developers ship code much faster, teams have hit a wall in pull requests and quality checks. CodeRabbit positions itself as the governance layer in between, a context-aware AI reviewer that understands a company’s codebase, flags bugs, security issues, and style problems, and drops feedback directly in IDEs, CLI, and *** platforms. The company, founded in early 2023 by Harjot Gill after he sold Netsil to Nutanix, has grown fast on that thesis. It announced a $60 million Series B led by Scale Venture Partners with participation from NVentures, Nvidia’s venture arm, plus CRV, Harmony Partners and others, bringing total funding to $88 million. At the time that round valued CodeRabbit at $550 million. Revenue is growing about 20% month-over-month and ARR has topped $15 million. More than 8,000 companies including Chegg, Groupon, Life360 and Mercury now use it, and over 100,000 open-source projects run it on GitHub Marketplace. Customers report big speedups, Groupon cut review-to-production time from 86 hours to 39 minutes. CodeRabbit says teams using it can cut human reviewers in half, and it’s doubling headcount to keep up with demand as vibe coding pushes more AI-generated code into production. - World Business News.

• 
  Egor (@jvqtil) reported 5 minutes ago

  New release of dpm - Decent Package Manager Now it supports installing packages from Direct URLs, besides GitHub and local files. Would really appreciate your feedback in DMs or GitHub issues!

• 
  VLAD (@VladConV) reported 9 minutes ago

  1/ Day 1 of building a portfolio project to break into Cloud/Backend roles. I'm an enterprise engineer (Java, .NET, SQL Server/Oracle) moving toward Cloud + DevOps. Decided the fastest way to actually learn is to build something real, the right way, not just watch tutorials. Here's what got done 🧵 2/ Started with the Domain layer — the innermost layer in Clean Architecture, zero external dependencies on purpose. BaseEntity, ValueObject, IDomainEvent as the building blocks. Then the first real entity: TodoItem. 3/ No anemic data bag. TodoItem has actual invariants: No public setters State only mutable through explicit methods (MarkComplete/Reopen) Raises a domain event on completion This is the difference between "a class with properties" and a domain model that protects its own rules. 4/ Test-first from the start: 16 unit tests covering those invariants before moving on to anything else. All green. 5/ Then the part that usually gets skipped in tutorials: real CI/CD. GitHub Actions running build+test on every push/PR. Branch protection requiring passing checks, no direct pushes allowed.

• 
  Brings (@bringsnft) reported 10 minutes ago

  Building Onervico is changing how I think about software. I’m not just writing code. I’m designing the idea, turning decisions into GitHub issues, and working with AI agents to build and review the product. One founder. AI as leverage. @OnervicoApp is the experiment.

• 
  Ted (@tedbrine) reported 11 minutes ago

  wait. Linear’s whole business model is just fixing github issues. they defiantly paid microsoft off to not improve the issue ux lol

• 
  Kyle Bradshaw (@SkylledDev) reported 12 minutes ago

  @bcherny I've got most of my tools setup to emit errors to a service that creates GitHub issues. Claude Code periodically checks the untriaged issues, investigates them, then comments a proposed fix. I approve or give feedback to the proposal, then another Claude creates a PR.

• 
  Rohan Paul (@rohanpaul_ai) reported 16 minutes ago

  New term “catastrophic remembering” CLAUDE.md has a ratchet problem: instructions are easy to add and increasingly hard to delete. Agent prompts may remember rules much longer than maintainers remember why those rules exist. Across 1,867 GitHub repositories, agentic instruction files more than tripled over their lifetime, with instruction count rising 226% on average. Deletion became less likely as instructions aged, and that decay was steeper in files touched by multiple human authors. The paper calls this “catastrophic remembering”: the instruction survives, but the reasoning that once justified it disappears. The proposed fix is a decades-old software practice: attach a comment recording the failure, hypothesis, and outcome behind each instruction, then hide that comment from the executing model. In the 51-step controlled IFEval setting, informative comments cut excess prompt size from +211.3% to +1.4%, with last-three-round constraint satisfaction at 44.0% in both arms. Comment-shaped noise did not reproduce the effect, pointing to preserved rationale rather than extra text. – arxiv. org/abs/2608.11095 Title: "Why Does CLAUDE.md Keep Growing? Catastrophic Remembering in Agentic Coding"

• 
  Stain (@pankostain) reported 20 minutes ago

  hai, i was originally going to post this on the github issues page but i prefer to ask to you directly: is there ANY possibility to update ginput so that it supports the original xbox mapping as an option for ppl want to play the game with the intended xbox controls? @__silent_

• 
  Chloe (@chloevalesquez) reported 22 minutes ago

  those go straight to a github issue instead of fighting luna on them. no point burning turns on a model that's not built for that kind of depth

• 
  Synapse Brief (@Synapse_Brief) reported 24 minutes ago

  Google's Python SDK just got a new string: gemini-3.7-flash. No model card, no pricing, no announcement. Just a name sitting in the GenAI SDK on GitHub, three weeks after 3.6 Flash shipped. That 3.6 Flash launch, for context, already cut output pricing from $9 to $7.50 per million tokens and trimmed output token usage by 17 percent across the board. If Google's iterating that fast on a model that just launched, the interesting question isn't whether 3.7 exists. It's why Flash keeps eating the roadmap while the model people actually asked for doesn't show up. Gemini 3.5 Pro was promised for June. It's now mid-August and Google's own model page still lists it as "coming soon." SemiAnalysis reported it's been quietly shelved in favor of a Gemini 4 pretraining run, Google has not confirmed that, and nobody outside DeepMind seems to know which is true. So you've got a flagship that's either delayed or dead, and a Flash tier iterating fast enough that 3.7 leaked in SDK code before 3.5 Pro even shipped. That's not a cadence problem. That's a company telling you where its actual confidence sits.

• 
  Brabalawuka (@brabalawuka) reported 26 minutes ago

  @thdxr Does the desktop beta support v2 server alr? I tried search with ChatGPT but give very vague answers based on github commit history / lissue log

• 
  martian (@martian0x80) reported 27 minutes ago

  github down again