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:

• 
  ©JΔΞOИ™✨ (@jasonshen_) reported 5 minutes ago

  Most voice agents recover by guessing. You cut in. It stops. Then it answers the sentence before yours. That’s not recovery. That’s a coin flip. StreamCore keeps the turn state. Duck -> Confirm -> Answer what you actually said. One Go server. Self-hosted. Apache 2.0. GitHub 👇

• 
  TypeDrifter (@typedrifter) reported 5 minutes ago

  @MozzyFX @dhh How is phishing different from a fan site ? That's the same issue with fake DMCA claims that plagues YouTube and mistakenly blocked github users.

• 
  Alice (@TrpedInWndrland) reported 6 minutes ago

  @night_owlll honestly I do think there is an issue where some devs will treat github as a downloading source and since github isn't designed for that it leads to this relationship

• 
  Stryker § (@strykertrades) reported 8 minutes ago

  Very suspicious, especially since the user is misleading people by showing a photoshopped GitHub repo while the actual file is hosted on some random third-party server.

• 
  jattinsran (@jattinsran) reported 10 minutes ago

  GitHub is down!!

• 
  Alexander Wolfheart (@AlexWolfheartt) reported 16 minutes ago

  @SynthPotato Here's a fun one: I bought the Xfx Mercury Magnetic Air 9070xt on launch day for 1050euros. In december last year, a local store had a special offer and I bought a Msi Ventus 3X for... 1050 euros. The performance difference was night and day, and yes, all my Radeons (got a 9070xt, a 7900xtx, and too many Amd-based handhelds) have had tons of issues, specifically driver-related, but also some hardware engineering incompetence (GCD Hotspot on the 7900xtx is in a moronic location, no wonder so many of them have that issue). I do, however, have an advice for you: Use DDU, nuke the drivers, remove Adrenalin. Reinstall gpu driver as standalone (driver-only). Do NOT get the full package, do NOT get Adrenalin. Instead, get Radeon Tuner off Github, and use that. Radeon Boost, Radeon Chill and Enhanced Sync keep them off, and many of your issues should disappear. Performance overall will still be mediocre, but at least you'll have less crashes and bugs :) My 9070xt (with which I got two 3DMark records) now sits on an Aaostar AG02 egpu dock, and hasnt been connected to anything since march this year :) Looks nice on my tech workshop room's table, though.

• 
  Andrew Gömez (@red_darkin) reported 18 minutes ago

  @Cyberadp Hi bro, i used opus 4.9 Related to the prompt i didn’t use an spwcific prompt i struggled for 2 hours trying to guide Claude for the correct path. But the initially prompt was “build a functional PoC based on this document” the document was the github issue about this CVE

• 
  Jerry Fragiskatos (@Aces1974) reported 18 minutes ago

  @Must_be_Ash Same thing happened to me. I was testing x402 services in a repo when an automated system shut down my account. It’s under review, and I’ve since moved to self-hosting. I don’t think GitHub is acting in bad faith, moderating at their scale is incredibly difficult. But in AI / crypto, days of downtime can equal months. It’s another reminder of the need for sovereign alternatives to centralized dependencies. @gitlawb is a great example @Must_be_Ash tag me when it’s back up what you built looks awesome 🙌

• 
  Vibranium Audits (@VibraniumAudits) reported 18 minutes ago

  What a real audit gets you: A named senior team reading your code line by line. No junior queue. Every finding with a fix. Reports public on GitHub. 480+ audits, zero exploits on secured code. Security isn't a PDF.

• 
  AGTP (@AGTPinsights) reported 19 minutes ago

  ChatGPT just expanded automation with website logins and app triggers today. Here's what you need to know. ChatGPT Work can now sign into password-protected websites on web and mobile, without OpenAI ever seeing your username or password. You can ask it to set up utilities for a new apartment, book a DMV or passport appointment, or check a reimbursement, basically anything you'd normally click through a browser yourself. OpenAI says the earlier version of this required a manual takeover step, where you logged in yourself inside a virtual browser and screenshots were disabled during that window. The new rollout brings direct sign-in support to both the web and cloud browser, still without exposing credentials to the model. Scheduled tasks also got an upgrade. Plus and Pro users can now set tasks to trigger off events in Slack, Gmail, and GitHub, like a new email or a pull request update, instead of only running on a fixed schedule. Free users are getting scheduled tasks for the first time too, capped at three tasks each. Users can also now share their favorite tasks for others to customize and use. Key numbers: - Free users: up to 3 scheduled tasks - Plus and Pro: event-triggered tasks on Slack, Gmail, GitHub - Login feature: rolling out today for Plus, Pro, and Business on web and mobile Both features are live now, and OpenAI says credentials are never seen or stored by ChatGPT.

• 
  Denson Ngumo (@denson_ngumo) reported 19 minutes ago

  @CloudStack Which release has a fix for the bug with reference #13233 on Github?

• 
  djSoniQ (@djsoniq) reported 24 minutes ago

  We'd have major data security issues with the way this NIN / NIMC thing is being used. Nigerians' PII just flying all over the place to anyone with a github account& enough tokens. Why are non-regulated sectors plugging into national identity infra in order to book kabu kabu???

• 
  Kavi AI Finance (@KaviFinance1) reported 24 minutes ago

  BUILDING HERMES AGENT FROM SCRATCH — PART 5 GIVING HERMES HANDS An agent becomes interesting when it can actually do things. That's where tools and MCP enter the architecture. When I first started connecting agents to tools, my thinking was simple: “If it can use the tool, give it access.” Great for demos. Not great for production. A CRM is a good example. Hermes might be able to: Read a customer Create a lead Edit a customer Delete a customer Those are completely different risk levels. So before connecting more tools, I make Hermes audit its own tool environment. Prompt: “Inspect my current Hermes installation. Do not modify anything. Find all available tools, MCP servers, integrations and permissions. For every tool report: 1. Name 2. Purpose 3. Available actions 4. Required credentials 5. Read/write behavior 6. Destructive actions 7. Potential failure or security risks Group everything into: READ WRITE DESTRUCTIVE FINANCIAL EXTERNAL COMMUNICATION Never expose API keys or secrets. Do not install or change anything. After the audit, recommend which tools should run automatically and which require human approval.” Then I review the result. Only after that do I connect the MCP server I actually need. And I start with a harmless READ operation. Not: “Delete everything.” The simple rule I use: READ → automate WRITE → validate DESTRUCTIVE / FINANCIAL / EXTERNAL → approval This is the part people miss about MCP. It isn't just a convenient way to connect tools. It becomes a bridge between your agent and the real world. CRM. GitHub. Databases. Calendars. Business APIs. Once Hermes has access to these systems, it stops being just a chatbot. It starts becoming an operator. But now we have a much harder problem: What happens when the tool works perfectly... and the agent makes the wrong decision? That's Part 6. Failure, verification and recovery. Bookmark this if you're building your own Hermes stack.

• 
  Kappaemme (@Kappaemme1926) reported 27 minutes ago

  MY CODEX SKILL JUST HIT 1,000 GITHUB STARS! I’m honestly so happy right now. I built First Customer Finder to help founders use Codex to find potential customers backed by real public signals. Seeing 1,000 people support something I made feels unreal. Thank you to everyone who starred it, tried it, shared it, opened an issue, or contributed. I’m going to keep making it better. What should I add next?

• 
  Neo Vector (@neovectormind) reported 29 minutes ago

  @FabianHedin Makes me wonder how many products are one GitHub policy change away from a multi-day outage.