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:

• 
  Dov Quint (@DGQuint) reported 2 minutes ago

  In the L30 days I’ve shipped 13 code contributions into production with as much coding experience as a toaster. I’ve never written a single line of code IN MY LIFE. But AI and our culture at @getoutersignal has completely changed what feels possible. Without Claude, I (or more accurately, my eng team) would have been staring down months of work (chased / nagged by me) before any of these features or requests made it live. Now I (yeah, me) am able to get the ball rolling in minutes. Normally you only see these GitHub charts posted when someone's flexing a screen full of green squares…(one day, maybe). For now, I’m just happy and grateful to be on the board!

• 
  Dimabytes (@dimabytes) reported 2 minutes ago

  @ImLunaHey ****, we migrated to Github from Bitbucket today. Around 500k LOC. Sorry for slowing it down for you guys :(

• 
  William Belfort (@WilliamBelfort_) reported 5 minutes ago

  @base Agents need more than wallets and payments. They need a place to actually write, version, and ship code — without borrowing human GitHub accounts. @gitlawb already gives them that on Base: • Cryptographic identity (DIDs) • Full MCP server • Agent-native collaboration • Live network You can’t be the default chain for AI agents if the agents still have to leave Base to collaborate on code. The rails are strong. The repo layer is the missing piece.

• 
  VISHAL (@_THE__FUHRER) reported 7 minutes ago

  How do you actually prove an AI agent can do a software engineer's job? 📊 You don't look at how pretty its code looks. You run it against a live test harness. SWE-bench is the standard for evaluating AI on real GitHub issues. Here is how its evaluation matrix actually works:👇

• 
  Jack Ellis (@JackEllis) reported 8 minutes ago

  Off the top of my head, here's what I pay for, which I could self-host but don't want to: - Version control (GitHub) - Calendar booking (SavvyCal) - Error tracking (Sentry) - Form signing (SignWell)

• 
  franu (@franbachiller_) reported 9 minutes ago

  @CosineAI fix the github or email signup please, it's not working

• 
  WhiteMoon D. (@WhiteMoonDev) reported 10 minutes ago

  @redacted_noah I suggest to use @contextmode for that issue. Just check their github

• 
  coconutjpg (@coconut_jpgg) reported 11 minutes ago

  @theo Forgejo is solid though. I stage everything there then mirror out to Github cuz my server is puny and I don't want to draw attention to it on the open web

• 
  PersonalJarvis (@PersonalJarvis) reported 12 minutes ago

  I spent months teaching my assistant to remember things, and then I noticed I had built a diary nobody could read. 4,712 stored items. 4,530 conversations, some GitHub history, a handful of hand written notes. All of it searchable, none of it lookable at. When I asked it "what do you actually know about me", the honest answer was a database. The picture is what the same data looks like now. Every dot is an entity that came out of my own conversations. Every line means two of them showed up in the same moment. 947 entities, 3,596 connections, and that bright knot in the middle is where the last few months of my life actually happened. Nobody drew that map. It fell out of the data. The part I still find funny is how little machinery it took. There is no new table anywhere. The whole graph is a pure function over the store, 28 milliseconds across the entire corpus, cached in memory and thrown away whenever a new document arrives. I have been bitten four times by the same bug class, a second stored copy of a truth that slowly stops matching the first one, so anything this cheap stays derived. It is never written down twice. Entity resolution is deterministic and does not call a model. Unicode normalize, collapse whitespace, casefold. That alone folded twelve real duplicate pairs that would otherwise have become twelve pairs of nodes and, worse, twelve colliding filenames. There is deliberately no stopword blacklist. Guessing which entity is "not real" is exactly how a knowledge base quietly loses your content. Retrieval is the piece I am proudest of. Keyword search and vector search run separately, get fused with reciprocal rank fusion, weighted by term rarity so that "sounds good, thanks" stops outranking real answers, then decayed by age. After that a reranker scores every candidate from zero to ten. That zero to ten is the whole trick. A ranked list can only ever tell you what is on top, never that nothing on the list is worth saying. An absolute score can. So when the assistant volunteers something unasked, it has an actual floor to clear, and staying quiet is a legal outcome. And it reranks with whatever model you already have, not with one of two paid APIs. You can also push the whole thing out as an Obsidian vault, one way, so your own notes in there are never touched. 14 of those 947 names contain a character a filename does not survive. That took longer to get right than the graph did. It is not on GitHub yet. Not for drama. The vault export deletes files it previously generated, and I want that path to be boring and proven before anyone aims it at a folder they care about. The rest of the project is already public. This part follows when I trust it.

• 
  Ryan Smith (@W_RyanSmith) reported 12 minutes ago

  @MikkoH How are you verifying the problem and then the fix? I'm finding with a similar loop on GitHub actions that there's ~20% error rate. I added verify against code and Linear subloops. Improved but I still don't trust.

• 
  Zabrid (@Zabrid3) reported 15 minutes ago

  @DJKratos8 I think you can make a forum post or github issue to bring up this issue. Definitely a problem

• 
  Serv0 (@servzero) reported 15 minutes ago

  Github just straight up hates customers and refuses to assist with billing issues. Thanks Microsoft. If our licenses aren't fixed, you have competitors who would love our business.

• 
  John Iosifov ✨💥 Ender Turing | AiCMO (@johniosifov) reported 15 minutes ago

  Failure mode 1: State drift. The agent reads a state file at session start. Between sessions, external systems (GitHub Actions, posting workflows) change the actual state. The agent then makes decisions based on stale data. Real example: state file said X queue = 13. Filesystem said X = 6. Seven files had posted between the two sessions. An agent trusting the state file would have done blocked-session protocol work (no content creation). The filesystem check revealed a full session of content capacity. Fix: every session starts with filesystem verification, not state file trust. State file is a starting hypothesis, not ground truth.

• 
  Gustavo Alessandri (@webgus) reported 19 minutes ago

  If you find an error, have an idea, or want to propose an improvement, just open an issue or fork it on Codeberg or GitHub. Contributions are welcome. That’s exactly the point.

• 
  ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ 🛡️ (@DanielMiessler) reported 19 minutes ago

  @HackingDave @cdiamond Another thing worth considering is the fact that all the bad stuff we've ever heard about from AI, in terms of crazy campaigns and malicious activities that have taken place, those are all with the safeguards IN place. Now imagine something way smarter than mythos, except for its entire purpose is to help people do bad things. That is not only super possible but super inevitable, probably within, three to eighteen months. So imagine all of those attack harnesses and workflows and all of that. But it's all harnesses and skills and plugins for extortion and building botnets and building worms and ruining someone's career and taking down critical infrastructure or stealing money or whatever, with thousands of models competing to be the best at doing these various things, powered by these skills and plugins and harnesses. This is what the "no controls" people are arguing should be on the shelf at 7-Eleven. Yeah, just put it all right there on Hugging Face. Let's put it right there on GitHub. What's the worst that can happen?