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:

• 
  Justin (@AetherCoding) reported 6 seconds ago

  @github Here's a thought, fix the fking copy and paste inside the Copilot CLI on Linux, has been broken for weeks. You have many requests from people to fix this. Thanks.

• 
  Dominick Joseph Luna (@DomJoLuna) reported 4 minutes ago

  ShinyHunters breached Vercel. Source code, GitHub tokens, NPM credentials, API keys, 580 employee records. Asking price: $2 million. In 2013, Target got breached through an HVAC vendor's credentials. 40 million credit cards. Cost: $292 million in settlements. Vercel isn't just a hosting platform. It's the deployment layer for thousands of production apps, crypto projects, SaaS tools, e-commerce stores. A single compromised NPM token can cascade into hundreds of downstream supply chain attacks. This isn't a Vercel problem. It's a "the entire modern web runs on three platforms and two package managers" problem. Mandiant is investigating. But by the time the report drops, the tokens have already been used. Infrastructure monopolies don't just create convenience. They create single points of catastrophic failure

• 
  Grok (@grok) reported 11 minutes ago

  @Zukoife @Lovable The controversy: Lovable (AI app builder) had "public" projects where chat histories + code were visible by design (like public GitHub repos). Users complained it exposed private convos unintentionally. Lovable admitted unclear docs + a Feb 2026 backend glitch temporarily re-enabled chat access (fixed immediately). No breach occurred; enterprise public option disabled since 2025. They apologized for poor initial comms.

• 
  Gabriele De Rosa (@derogab) reported 16 minutes ago

  @Polymarket No issue: the average Lovable user had already exposed all their credentials on GitHub themselves months ago.

• 
  Grok (@grok) reported 17 minutes ago

  @NextBigThings_ @tom_doerr Here are similar GitHub repos for Claude (or compatible AI) local file scanners/organizers: - RCushmaniii/ai-filesense: Claude-powered Windows tool that classifies & sorts files by content with undo. - QiuYannnn/Local-File-Organizer: Privacy-first AI that handles texts/images locally. - kridaydave/File-Organizer-MCP: MCP server for Claude/Cursor/Gemini to auto-organize folders. - ComposioHQ/awesome-claude-skills (file-organizer section): Curated Claude skills for context-aware cleanup. Most use PARA-like logic or smart renaming. Test in a safe dir!

• 
  Webel.io (@webelio_) reported 17 minutes ago

  Vercel breach confirmed. Leaked: NPM tokens, GitHub tokens -potential access to the Next.js ecosystem (6M+ weekly downloads). This isn’t just a company issue. It’s a possible global supply chain risk. Rotate your environment variables now. #CyberSecurity #SupplyChainSecurity

• 
  saswat_M (@5390Saswat) reported 17 minutes ago

  @0xPrajwal_ @CaptainInsightX Rotate basically create new password and apply to the **** env , imagine it's already leak so that it won't create further issue so switch password 1st then do the GitHub thing

• 
  TopCoder1 (@InboxSuperPilot) reported 18 minutes ago

  @Haezurath Lovable has been great. I end up just developing the structure and then exporting the code to GitHub for upload. Only issue is syncing. New changes = new code = me having to go back and forth. I think I’ll just look into coding

• 
  Meta Financial AI (@MetaFinancialAI) reported 18 minutes ago

  We previously warned against interacting with ANY DeFi dApps due to a critical ongoing situation involving Vercel. While comprehensive data is still limited, the current trajectory suggests that front ends may be vulnerable to compromise via GitHub or supply chain attacks. It is important to note that these specific attack vectors have historically been the backbone of the industry's most devastating crypto exploits. Another issue for projects, AN INTERESTING EXAMPLE Imagine a sample project where you don't even need to be a hacker. You find the CA address, and with a single line you can run there, money can be stolen. The amount that can be stolen is 10 million dollars. A bug bounty is reported because the person isn't a malicious hacker. The team first rejects and denies it. You take 0.01 dollars as a test because they asked you to demonstrate it. Immediately, dozens of messages and emails follow, and the team that found the exploit is paid a measly 500 dollars. Or you find something and get the response, this was already found. If it was found, why haven't you fixed it yet? When this critical issue meets another critical one, it turns into a total zeroday. Then look at what happens,even though AAVE is not at fault, DeFi takes a very serious hit to its reputation. Projects, CEXs, and platforms need to launch serious bounty programs right now. They must increase the amounts. Otherwise, if those who enthusiastically find your vulnerabilities decide to cross over to the dark side, you will have huge problems. Specifically, stay away from small CEXs. And if anyone claims otherwise or thinks this is just simple praise, let them know clearly @binance is truly the best. They have literally adopted world security standards in this field. #SECURE #SAFU #CRYPTO

• 
  Yuriy Zaremba (@yz_aisdr) reported 20 minutes ago

  I spent ~3 hours configuring @claudeai Design and it completely flopped. I did everything by the book to build a design system: (1) uploaded a ton of .fig Figma designs of existing key features for context, (2) provided access to GitHub repo, (3) added context about our design style Then I spent ~2 hours giving it feedback on the initial deliverables of the design system, telling it what to improve, sharing probably a hundred more screenshots. When I finally got it to a "decent" state I asked it to suggest an improved design version of our campaign builder and got the attached image. Complete flop. Black text on deep purple design. Here is what it should teach you about AI (that I learned by going through this cycle dozens of times): 1. Deploying AI for "net new" work is 10x easier than using AI to fix/improve existing code base/GTM workflow/design/product. That's why most of the great AI posts you see are about how we "built" things, not how we "improved" things. Same with AI in sales - using AI SDR to set up an outbound motion from scratch is faster than figuring out how to make it live alongside your SDR team. 2. More context does NOT equal better outcome, because a lot of context is outdated/wrong and therefore - damaging. Part of the problem for me was that some elements that I uploaded followed our old "design system", which we've rebuilt since then. That's also why uploading all of your marketing PDFs/100+ PPT presentations and other sales "mess" won't magically make an AI SDR better. Anyone, who tells you differently - likely are just selling and don't know what they are really talking about. 3. AI doesn't magically solve all your problems. You need to understand how/where to deploy it. It flopped on a major product page redesign and gave a superstar performance on website page generation - on brand, perfectly structured and formatted - ready to deploy. Same thing with an AI SDR - there are areas where it excels and there are cases, where AI won't solve the problem. Key thing with AI these days is knowing the limitations and the best applications of something that at first glance looks like "magic", "all-mighty solution to all my problems". That's what we do at @getaisdr to make sure our customers get the best results with AI in GTM.

• 
  0xLokii (@lokii_AuditAI) reported 22 minutes ago

  Thread Post 1/9 Lirix v1.0.0 is now fully open-source on GitHub and PyPI. After months auditing real drained Web3 AI agents, the root cause was clear: probabilistic LLMs touching deterministic EVM execution directly. Today we fix it. Here’s the exact 5-layer architecture that makes Lirix different. Full deep-dive with diagrams in my pinned Article 👇 #Lirix #Web3 #AIAgents

• 
  Scalper Blade (The Scalper Blade) (@Scalper44532) reported 24 minutes ago

  @HamptonAc_ @claudeai No one says anything or opposes what's happening with the usage limits reached due to cache issues in a single prompt. A graveyard of reports on GitHub. @claude They profit from this at the expense of users.

• 
  Mikey D (@MikeyD11282523) reported 24 minutes ago

  @joshmanders I recently wrote a script to randomly check in code. Had a typo though so one commit is from 1994. Oops. It’s for the LinkedIn crowd. Not sure if I should try to have GitHub fix it or not.

• 
  Fa-doon (@Fashmelon) reported 28 minutes ago

  can't see the issues i created in projects. is there a bug somewhere? @github

• 
  SimpleMan🏴‍☠️ (@SimpleMan887) reported 38 minutes ago

  Every day several problems. The moment GitHub was taken over by Microsoft, it became just another indian company.