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:

• 
  htrowii (@htrowii) reported 1 minute ago

  @brainage19 i set my flake up with copy pasting github dotfiles on bare metal it was terrible

• 
  catman (@catmanyau) reported 3 minutes ago

  @CricTalk29 for me, losing Cursor would hurt most because it sits directly in the editing loop. would the vote change if github outages were limited to code hosting but issues and reviews stayed available?

• 
  small_j (@a_small_j) reported 11 minutes ago

  @smalldocs_org recently crossed 200 stars on GitHub and 20 forks. SmallDocs is the first open source project I've managed. Handling other people's pull requests is not easy (and I need to improve). They implement features you're not considering and fix bugs you didn't know you had. Extremely useful, but if you're squeezed for time and trying to develop core functionality, it's hard to manage both things well.

• 
  OverlyPositivePatriot (@JBrowsing2023) reported 17 minutes ago

  As a IT professional, I have a recommendation @github should take seriosuly. We should only get a notifican from Github when it is up rather than when it is down. Reliability is a disaster for this product.

• 
  Jason Waters (@jasonwaters87) reported 22 minutes ago

  Anthropic just open sourced the code Shopify runs their shopping agent on. Free on GitHub. And I’m having lunch with a surgeon in San Jose last month and he tells me a patient no shows and nothing happens. Nothing. He has to walk up to the front desk himself and ask “did you guys call them?” 3 or 4 schedulers looking after 75 doctors. One automated call before the appointment and that’s it. His own dermatologist sends him three reminders. He called that “an easy fix.” That’s a merchant agent. Reminds the patient, rebooks the no show, tells him Thursday isn’t full so he can put a surgery on it. The code is sitting there free. Somebody still has to walk into his office and build it.

• 
  Dr Milan Milanović (@milan_milanovic) reported 24 minutes ago

  How Cursor made *** scalable The thing with *** is that it never was designed to be scalable. Your repo lives on the disk, and *** client expect every read to be consistent. This was a problem on GitHub, where shared filesystems and replicated storage failed before 2013. The GitHub built 𝗦𝗽𝗼𝗸𝗲𝘀, and it became the industry standard. This means that every repo is stored as three full copies on three servers, and every push runs a vote (three phase commit). A majority of servers must confirm before it exists. This works, but with high cost, because every push is slow as the slowest server. When we add new servers, it makes it even slower. Now Cursor took some opposite direction with 𝗖𝗼𝗻𝘁𝗶𝗻𝘂𝗶𝘁𝘆. The repo history is now written as a log in S3, and this is only source of truth. Any push counts only if it is located in the log. The servers don't need to keep anything important, they are just cache. Any server can take a push, and idle repos are dropped from disk and rebuilt from the log when it is needed. This resulted in 120 pushes per second on standard S3, and over 300 on S3 Express. Their tests have shown that read capacity grew linearly up to 100 replicas. Why is this important now? Because of AI agents mostly. We now have more code, PRs, CI runs and many small repos. All of these repos would need three full copies in the old model. This means that we achieve scale by removing parts, not adding them.

• 
  R 'Nearest' Nabors (they/them) (@rachelnabors) reported 28 minutes ago

  @Paul_Kinlan Honestly, the linear method helps. Think of it as having a never-ending trough of issues that agents can pull from. I don't even use linear. I just use GitHub with linear flavouring added

• 
  Blue Collar Executive (@A_Sober_Drunk) reported 28 minutes ago

  on the third try at the same problem, I told Grok to "stop and go search stack overflow or github or something"... five seconds later... Literally the exact issue, problem solved. That's how new global rules are born.

• 
  Spectra☢️ (@Spectra010s) reported 32 minutes ago

  @izzyCodes_ and you too Chief Check GitHub issues

• 
  dug_vt (@dug_vt) reported 32 minutes ago

  @sonemic rym users don’t use spotify they download flacs off soulseek and transfer them to a server connected to their pc and play them from a self hosted music player from github

• 
  Kevin Minnelli (@minnelli) reported 34 minutes ago

  WTF - Grok Bot can't fire on schedule to save it's life. The scheduled routines are just broken and at best unreliable. I want to love this product. When you set the cron job it doesn't work. It tells you try Cloudflare, sure let's set that up and burn tokens, then that doesn't fire to wake them. Oh, let's try GitHub now and use that....all failed. I had to wake it again this morning before the market opened. Anyone else feeling frustration in this regard?

• 
  moledao (@moledao_io) reported 37 minutes ago

  Web3 Remote Job Scams: A 2026 Field Guide Introduction Over the past few years, Web3 has come to represent a new world of opportunity for many young people. New roles, new narratives, and new stories of wealth have inspired countless people to enter the industry with high expectations. Remote work, stablecoin-based compensation, and a greater emphasis on ability than academic credentials can be especially attractive to professionals at the beginning of their careers. As a recruitment platform that works with job openings and candidates every day, however, we have also seen the other side of the industry. A significant share of supposed recruitment activity is not recruitment at all. It is fraud disguised as hiring, designed to steal the funds in job seekers’ wallets. According to Chainalysis’ 2026 report, cryptocurrency scams and fraud caused an estimated $17 billion in losses worldwide in 2025. Impersonation-related attacks increased by 1,400% year over year. Fake recruitment is one of the most common ways impersonation and social engineering are being applied to job seekers. What Happened to Us This month, it happened to us. We were contacted through Telegram by someone claiming to represent a US-registered technology company. They said the company urgently needed to hire Web3 engineers and wanted our support in sourcing candidates. After further investigation, we were unable to verify whether this person had actually been authorized by the company. We also could not rule out the possibility that they were impersonating a legitimate business. To avoid causing further harm to an organization that may itself have been a victim of impersonation, we will not disclose the company’s full name. At first, there were almost no obvious warning signs. The contact provided a business registration document and a polished company profile. Interviews were scheduled through Calendly, job openings were hosted on Ashby, and meetings took place over Zoom. These are all professional tools commonly used by legitimate companies, making it easy to assume that a company using them must be trustworthy. In reality, forging a registration document and creating Calendly or Ashby accounts require very little effort. Almost anyone can create the appearance of professionalism at minimal cost. This experience taught us an important lesson: the legitimacy of the tools surrounding a hiring process tells you very little about the legitimacy of the company behind it. The details of how the people involved behave are far more revealing. The partnership also progressed with unusual ease. All contractual documents arrived at once and appeared ready to sign. There was no friction at any stage. When discussing the recruitment fee, we initially proposed 15%, which the other party immediately accepted. We then tested an increase to 20%, and they accepted again without hesitation. Anyone with experience in recruitment delivery knows that fees are often one of the most difficult parts of a headhunting agreement. Clients may negotiate repeatedly over a difference of just two percentage points. The pace was also deliberately compressed. Interviews were often scheduled only one or two hours in advance, leaving almost no time for verification. Once confirmed, meetings were then repeatedly cancelled or rescheduled due to supposed last-minute conflicts. Several additional warning signs gradually appeared. The registration documents looked legitimate at first glance. Upon closer comparison, however, the names of the people listed in them did not match the information we were able to verify independently. The contact also made an unusual request. While verifying whether candidates were currently employed, they asked us to find out whether those candidates used LinkedIn frequently. Normal employment verification does not require this information. A person’s activity on a professional networking platform primarily reveals whether they have an accessible network that could quickly be used to verify their identity, employment history, or recent activity. At the same time, the contact prohibited us from sourcing candidates through LinkedIn or Telegram. They claimed their internal team was already using those channels and wanted to avoid duplicate candidates. In practice, this restriction pushed external recruitment partners into channels where independent cross-checking was much more difficult. The geographic requirement was even harder to explain. A US company was willing to consider only Chinese-speaking candidates and applied unusually strict screening standards that appeared unrelated to technical ability. In retrospect, we suspect that the screening criteria may have favored candidates who were easier to persuade and more likely to have higher incomes or larger asset balances. We cannot, however, confirm the group’s true intentions. Each of these warning signs could have been rationalized on its own. A client may have unusual preferences, legitimate concerns, or simply an unprofessional hiring process. It was only when the signals were considered together that the larger pattern became visible. Then came the interviews. Candidates joined the meetings, but the interviewers asked no questions about their project experience or technical background. Instead, they immediately provided a GitHub repository and instructed candidates to clone it onto their local machines and run it. The repository involved encryption and signing operations using cryptocurrency wallet private keys. Before the interviews, we had explicitly asked whether candidates would need to download or run anything. The contact told us they would not. Once the meetings began, however, the candidates received the exact opposite instruction. One candidate offered to share his screen, inspect the code locally, and walk the interviewer through it line by line. The interviewer refused and insisted that he download and run the repository on his own computer. The candidate ended the meeting. Other candidates quickly noticed that something was wrong and stopped as well. The most carefully designed part of the operation was not what happened during the interviews, but the feedback that followed. If a candidate ran the code, the interviewer gave positive feedback, said the candidate had performed well, and advanced them to the next round. If a candidate remained cautious and refused to run it, the interviewer told us that the candidate had falsified their résumé and instructed us to blacklist them immediately. That second response was not merely feedback. It was an instruction designed to prevent further communication between us and the candidate while allowing the wider operation to continue. This is something we hope every recruitment professional remembers: when a client asks you to blacklist a candidate without providing credible evidence, the request may reveal more about the client than it does about the candidate. Another common feature of these operations is that they do not need to interview every candidate. They only need a small number of people who are willing to execute the code. As a result, the hiring process will often stall abruptly once enough potential targets have been identified. What We Did Afterwards We immediately terminated all cooperation with the contact and removed the related job listings. We contacted every candidate who had entered the process to determine whether anyone had downloaded or executed the code. We also provided guidance on device inspection, credential rotation, and wallet security. All contracts, chat histories, meeting information, repository URLs, and account details have been preserved. Reports have been submitted to Telegram, GitHub, Ashby, and Calendly. At the procedural level, we have rewritten our client identity-verification process. Going forward, we will not accept recruitment assignments or partnerships without conducting independent callback verification through a channel the contact does not control. Candidate security briefings will also become a standard step before we introduce anyone to a client. Other Recruitment Scams Currently in Circulation What we encountered was only one variation. Several other methods remain active and deserve close attention. Malicious Take-Home Assignments Malicious interview assignments are currently one of the most widespread forms of recruitment-related attacks. Attackers impersonate recruiters or hiring managers on LinkedIn, X, or Telegram. They advertise senior roles with compensation well above market rates and frequently target professionals working with React, Next.js, Solidity, and blockchain technologies. Candidates are then given a technical assessment in the form of an npm project or GitHub repository and instructed to run it locally. Unit 42, the threat-intelligence team at Palo Alto Networks, refers to this activity as “Contagious Interview” and tracks it under the identifier CL-STA-240. The campaign was first publicly documented in November 2023 and has been linked to North Korea–associated threat actors. The malware used in these campaigns includes BeaverTail and InvisibleFerret. These cross-platform payloads target Windows, Linux, and macOS devices and are designed to steal sensitive browser information and cryptocurrency wallet data. According to security researchers, more than 197 malicious npm packages associated with this attack path have been distributed since October 10, 2025, accumulating more than 31,000 downloads. Common warning signs include recently created repositories, abnormal commit histories, and contributors whose identities cannot be verified. These are only indicators, however. Attackers can compromise established accounts, fork legitimate long-running repositories, or manufacture months of commit history in advance. No single signal can prove that a repository is either safe or malicious. Fake Meeting Software Fake meeting applications are another major threat. Attackers approach targets with an investment opportunity, partnership proposal, or interview invitation. Shortly before the meeting, they claim that Zoom is not working or that the company uses a different conferencing platform. The target is then directed to download the supposed meeting software from a specific website. Cado Security Labs has tracked one such campaign, known as “Meeten,” since September 2024. The campaign distributes a cross-platform information stealer called Realst. The group uses AI-generated company profiles to make its operations appear more credible. The names and branding of its meeting applications change frequently, with known examples including Clusee, Cuesee, Meetone, and Meetio. The malware targets cryptocurrency wallets and Telegram credentials, as well as iCloud Keychain data, banking information, and browser cookies. The solution is not to memorize an approved list of meeting applications, since legitimate companies may use many different tools. The safer rule is never to download meeting software from an unfamiliar domain sent directly by an interviewer. Download the software independently from its official website or an official application store, and verify the domain carefully. Malicious Offer Files Fake offer documents are another common attack method. In March 2022, attackers stole approximately $540 million from Axie Infinity’s Ronin Bridge, although later reporting placed the total value closer to $625 million. Subsequent investigations found that the initial point of entry was a fraudulent job offer delivered as a PDF. A senior engineer at Sky Mavis was contacted on LinkedIn by accounts impersonating another company. After completing several rounds of interviews, the engineer received an extremely attractive offer in PDF format and downloaded it. That file introduced spyware into the system. The attackers eventually gained control of five of the network’s nine validator nodes. Sky Mavis confirmed that an employee had been targeted through social engineering. In April of that year, the US Treasury attributed the attack to the Lazarus Group. What makes this case especially significant is that the victim was a senior engineer at the company that was ultimately compromised, and the attack was supported by a complete, multi-stage interview process. The final payload was simply a file. Terminal Paste Attacks Attacks that instruct victims to paste commands into a terminal or system run box have grown rapidly over the past two years. They are commonly known as ClickFix attacks. During an interview or onboarding process, a page may claim that the user’s browser has encountered an error, that their identity must be verified, or that a system component needs to be repaired. The page then provides a command and instructs the user to paste it into a terminal or run dialog. In May 2026, Microsoft disclosed a campaign targeting macOS users through lures disguised as system utilities. The campaign was used to distribute information-stealing malware. Other security companies have identified similar samples containing asset-transfer functionality. The malware first checks whether a wallet contains funds and then transfers those assets to an address controlled by the attacker. The rule here is simple: no legitimate recruitment process requires you to paste a command you do not understand into your terminal. Not once. Deepfake Interviewers Deepfake interviewers have already begun to appear. Real-time face-swapping technology is now advanced enough to support an apparently coherent interview. The person on screen may appear to be a senior executive from a well-known company, speak professionally, and have a verifiable public résumé. There are two practical ways to respond. First, ask the person to perform an unexpected physical action, such as briefly covering half of their face with their hand or turning their head 90 degrees to the side. Current real-time face-swapping systems may still reveal visual inconsistencies when the face is obstructed or shown from an extreme angle. Second, conduct an independent callback using contact information published on the company’s official website. This remains the most effective method of verification. Malicious Wallet Signatures A wallet-signature attack does not require your seed phrase. The interviewer may ask you to test a product, review a dApp, claim an onboarding airdrop, or complete an onchain identity-verification step. You are then instructed to connect your wallet and sign a transaction or message. Certain signatures or malicious transactions can give an attacker permission to transfer your assets. The level of risk depends on whether you are signing a basic message, a Permit, a token approval, or an onchain transaction. If you do not understand exactly what a signature authorizes, do not approve it. The boundary should be clear: no interview or onboarding process requires you to connect a personal wallet. A recruiter or employer has no legitimate reason to require a job candidate to perform an onchain transaction. Upfront Fees and Identity Misuse Upfront fees and identity misuse are among the oldest recruitment scams, yet they are still frequently overlooked. The first typically involves demands for a security deposit, training fee, or equipment payment before employment begins. The second asks a candidate to use their identity to register an account with a cryptocurrency exchange or open a bank account. This can carry consequences far more serious than financial loss. If the account is later used to process criminal proceeds, the person whose identity was used may face criminal liability. Any request for payment before employment is a red line. If someone asks you to register an account, receive funds, or move money on their behalf using your own identity, end the conversation immediately. How to Protect Yourself Before an interview, take three low-cost precautions. First, verify that the company genuinely exists. Review its official website and registration information, then examine whether the online histories of its team members are consistent across different platforms. Check whether the names listed in corporate documents match publicly available information. Second, independently contact the company through a channel the recruiter cannot control. Use an email address or phone number published on the official website. Do not use contact details provided by the person approaching you. Third, pay attention to two recurring warning signs: interviews scheduled only one or two hours in advance and then repeatedly changed, and compensation that is clearly above the market rate for the role. Both patterns appear in a large number of reported cases. During an interview, pause whenever you are asked to download, install, or run anything. You may offer to explain your approach over screen sharing, but remember that screen sharing itself does not provide protection. If code is running on your own machine, you can still be compromised even if you show the interviewer every line beforehand. The interviewer’s response is often more revealing than the request itself. If they refuse to explain what the code does, refuse to provide an isolated environment, or insist that you run it on a device containing your wallets and work credentials, end the interview immediately. When sharing your screen, share only the specific application window required—not your entire desktop. If the first interview contains no questions about your experience, projects, or technical background and moves directly to running code, you have every reason to end the call. We do not recommend that job seekers attempt to run untrusted code themselves. If analysis is genuinely necessary, it should be handled by someone with appropriate security expertise inside a disposable, isolated virtual machine that contains no credentials, does not mount directories from the host system, and has restricted network access. A container is not a purpose-built malware sandbox. Misconfigured directory mounts, permissions, or network access can still expose the host environment. Ordinary job seekers should not attempt this on their own. If you are a recruiter or regularly recommend opportunities to other people, incorporate these warnings into your standard process. Before introducing a candidate to a client, clearly tell them not to download unfamiliar software or browser extensions, not to run unknown code or scripts, and to share only the necessary application window during screen sharing. If a client asks you to blacklist a candidate without credible evidence, contact the candidate directly and verify what happened before taking action. Conclusion The crypto industry has spent years removing trust from transactions. You do not need to trust the counterparty because there is a contract. You do not need to give anyone your private key because it remains in your possession. At the protocol level, the industry has solved this problem remarkably well. Recruitment places people back in a much more primitive position. A stranger claims to be someone, and you must decide whether to believe them. Business registration documents and meeting links can be forged. Even an interviewer’s face can now be replaced in real time. Not a single dollar of the more than $500 million stolen in the Ronin attack was taken through a flaw in cryptography. What failed was the human layer. Every method described in this article relies on the same force: speed. The opportunity may disappear. Other people are competing for it. You have to act now. In an industry where everyone is urging you to move faster, giving yourself permission to slow down may be your most effective line of defence. If you have encountered a similar approach or recruitment process, please let us know. We hope this article helps more people recognize the warning signs before it is too late.

• 
  Lummox (@Lummox_eth) reported 39 minutes ago

  My own built Grok Bot turned $1,000 into $5,300 for last 17 hours. Now the project behind it is sitting around $25K market cap. We already pushed past $60K once and gonna hit $200k soon The bot is still running. The utility is almost ready. GitHub is live. Dev tokens are burned. I’m still buying. Nothing about the actual project changed because the chart went down. At $20K MC, this is the entry I personally like far more than chasing the first move. The target hasn’t changed either. $100K+ is where I want to take this next. $LUM is just getting started.

• 
  The Startup Ideas Podcast (SIP) 🧃 (@startupideaspod) reported 39 minutes ago

  One of the best skills to install right now is my friend Peter Yang's no AI slop skill. It's an editor. It hunts for the patterns that make writing feel AI generated and strips them out, while trying to preserve your actual voice. The second part is the hard one. Most writing tools make you cleaner and sand off the interesting parts, so everyone ends up sounding the same. You already know the smell. The grammar is fine, the syntax is fine, and it still reads like a keynote from a fake SaaS conference. It writes "it's not x but it's y." It uses "quietly" a lot. Here's how I run it: 1) Install it: npx skills add, then the GitHub link. 2) Write a rough draft yourself. An outline is fine, messy is fine. 3) Get your real points down, the ones only you would make 4) Ask the skill to remove the AI patterns and keep your voice. Step 4 only works if step 2 is real. If you ask AI to write the whole thing, there's no voice left to preserve. If you're building products, you're writing constantly. Tweets, landing pages, cold emails, launch posts, product updates, onboarding copy, investor updates. Nobody replies to say "this was written by AI." They just trust you less and keep scrolling. Write the messy draft, run the skill, then post it.

• 
  Ifebuche Omeke (@omeke_NC) reported 39 minutes ago

  Providing compute, storage, networking and managed services in the cloud. Terraform. Bicep. CloudFormation. Pulumi. They all solve the same problem: Defining and provisioning infrastructure as code. GitHub Actions. Azure DevOps. GitLab CI. Jenkins.