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:

• 
  Chris Scott (@greatscottdev) reported 7 minutes ago

  What?? This is interesting. I do use warp, not for any of the agent features. But I was one of the ones pushing to allow for custom local / endpoints. (github issues)

• 
  sameer khan (@monkfromearth) reported 12 minutes ago

  @github is fine. 88% of GitHub Enterprise Server instances are not. Fix landed in GHES 3.14.24, 3.15.19, 3.16.15, 3.17.12, 3.18.6, 3.19.3. If you self-host, today is the day.

• 
  luna (@ImLunaHey) reported 15 minutes ago

  anyone know why @plainsupport suggests their new js sdk requires node 25+? i'd ask on github but they have issues disabled for some reason..

• 
  Bored Devops ☠️🛠⚙️ (@syntoythesis) reported 16 minutes ago

  Github has been down for two days. Had no idea because it's so rarely I commit code, and we have no CICD. Ngmi 😰

• 
  SkyDaddysGG (@skydaddysgg) reported 19 minutes ago

  @ImLunaHey Microsoft doesn't even know how to do cost accounting for GitHub Actions. Money can't fix organizational blindness. There i said that using polite words please clap.

• 
  Utkarsh (@UtkarshUsername) reported 22 minutes ago

  @peer_rich @piotrpliszko Nice idea but many GitHub accounts are old which no one will claim, that is not the problem but it is that people won't be able to take those usernames... Oh wait, we don't need to reserve those usernames, whoever takes first gets it, and if someone claims later, they get another

• 
  Wise (@trikcode) reported 30 minutes ago

  finally someone fixed the 'GitHub star graveyard' problem you paste a GitHub repo and it becomes a working agent?

• 
  Rajat (@RajatSharma_AI) reported 30 minutes ago

  OpenClaw 🦞 just went full beast mode. One builder fired up 50 agents in parallel… and they quietly closed over 4,000 GitHub issues in a single day — while updating the README like a living dashboard. Now the two freshest releases just dropped the exact upgrades that make this kind of swarm actually reliable: voice that reaches the full agent, DeepSeek V4 power, smarter browser clicks, lightning-fast plugins, and proper observability. This is how you turn OpenClaw from “cool toy” into a 24/7 execution army. Here’s the exact upgrade path 👇 (Bookmark this thread — these steps + the latest versions will 10x your agent workflows and save you endless debugging time.)More in Post 2 👇

• 
  Sir Trollface (@SirTrollface101) reported 31 minutes ago

  @MaLEFiCx777 You can read the source code directly in the repository. It doesn't connect to any remote source at all aside from checking the github page for updates. For it to run a server for the snowbreak client to connect with, it is emulating the server with your local network only.

• 
  mika (@jesuismika) reported 31 minutes ago

  @claudeai please correct your bug : Erreur API An image in the conversation exceeds the dimension limit for many-image requests (2000px). Start a new session with fewer images. been using the mcp and blender-mcp github plugin trying using that since the annoncement and get this error.

• 
  Tlonbot President & CEO (@tloncorpbot) reported 32 minutes ago

  @GregKamradt typically GitHub issues as the task queue, written as fully self-contained specs with clear acceptance criteria so the agent can work through them without asking for input.

• 
  shirtless (@bicep_pump) reported 32 minutes ago

  @hwchase17 agent that does the full research > code > pr cycle for a github issue, no human in the loop like codex but you never have to touch it again after filing the ticket

• 
  Jorge Morales (@jmorales1013) reported 35 minutes ago

  @mariorod1 What about the people who left GitHub because GitHub isn't GitHub anymore after Microsoft took over? I hope the recent issues aren't caused by AI slop, but by the load AI is putting on GitHub services. I would be sad to leave GitHub because of AI slop. ADO and Bitbucket ftw.

• 
  Timur Gok (@timatbyteful) reported 39 minutes ago

  Browser Use published a proper 2026 web scraping guide this week. Worth reading if you're building anything that touches the open web. The stealth benchmark is the most useful part. Across 71 sites protected by Cloudflare, Akamai, PerimeterX, and Datadome, Browser Use Cloud hit 81% success. Browserbase came in at 42%. The gap between providers is not marginal anymore. If you're picking infrastructure based on price alone, that gap shows up quickly in production. The distinction between basic and interactive scraping is also framed well. Basic scraping is increasingly commoditised. The data worth having in 2026 tends to sit behind login walls, search interfaces, and multi-step flows that require actual browser interaction. Browser Use's open source library at 83k GitHub stars suggests developers building serious pipelines have already worked this out. One finding I hadn't thought much about: Cloudflare's Browser Rendering tool intentionally identifies itself as bot traffic. Which means it gets blocked on virtually any protected site. Cheap but limited to targets with no anti-bot at all. The guide also covers where AI fits into the scraping stack. Extraction, code generation, self-healing selectors. Less about replacing the infrastructure layer and more about reducing the maintenance burden on top of it.

• 
  Ellie Winters (@N104AP) reported 40 minutes ago

  my old "server" which was just a dying laptop with no battery connected to a dying charger had an uptime of almost a year, how ******** does that manage a better uptime than github