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:

• 
  The Jobfather ® 🇯🇲🇨🇦🇬🇧 (@TheJobfather__) reported 1 minute ago

  A junior developer should not hide all their work behind GitHub. GitHub matters, but hiring teams need context. What problem did the project solve? Who would use it? What decisions did you make? What would you improve next? A portfolio site helps organize the proof so the reviewer does not have to investigate your value like a detective.

• 
  Kracked Devs (@KrackedDevs) reported 4 minutes ago

  India’s takedown of BitChat accidentally turned into a live demo of why decentralized tools are so hard to kill. The government can force GitHub to delete repos because it’s a single company with servers and lawyers. BitChat’s code was quickly mirrored onto a peer to peer code network, where there’s no central server or company to order around. The app itself already works without internet, SIM, or phone numbers, using Bluetooth mesh so there’s no central point to shut down. Once people clone and “seed” the code, every new copy is another place the government’s order can’t reach.

• 
  Open-source Projects (@the_osps) reported 4 minutes ago

  • Terminal-native development with no context switching required • GitHub integration for repositories, issues, and pull requests via natural language • MCP-powered extensibility with custom MCP server support

• 
  Louis Maddox (@permutans) reported 5 minutes ago

  Maybe they expect noone to report it bc it's assumed CC supersedes it but Anthropic have broken in-chat GitHub sync entirely now (trying to use it just appends the repo URL to your prompt, then it fails to crawl the URL in the chat and guesses what was in it)

• 
  KrunalSinh Sisodia (@krunalbuilds) reported 8 minutes ago

  1/ An empty README = red flag. It tells me you built it and forgot it. Write one sentence. Just one. 2/ Committing directly to main. Every. Single. Time. Branches exist. Use them. 3/ 47 repos, 0 pinned. You're hiding your best work behind your worst. 4/ Commit messages that say "fix" or "update." Fix WHAT? Update WHAT? 5/ No contributions to anything public. Forks count. PRs count. Show you exist. 6/ Projects with no live link and no screenshots. I'm not cloning your repo to see if it works. 7/ A bio that says "Aspiring developer." You're a developer. Own it. 8/ Last commit: 8 months ago. Even one push a week signals you're still alive. Your GitHub isn't a storage drive. It's your pitch deck. What's the one thing you're fixing on yours today? Drop it below 👇

• 
  Berzeck (@Berzeck5) reported 10 minutes ago

  Broadly speaking, Open source is not merely an ideological preference. It is one of the most powerful mechanisms for accelerating innovation, creating real competition, and preventing technological control from becoming concentrated in a handful of companies. Microsoft learned this lesson the hard way. Steve Ballmer once called Linux a “cancer.” Later, during the SCO v. IBM litigation, Microsoft paid SCO substantial licensing fees and helped introduce it to BayStar, which participated in a $50 million investment supporting SCO while it was attacking Linux, this connections was strong enough that many reasonably interpreted it as an attempt to slow Linux adoption through indirect legal pressure. It backfired spectacularly. SCO’s central claims collapsed, the company went bankrupt, and Linux continued expanding until it became dominant across servers, cloud infrastructure, and supercomputing (500 of 500 most powerful super computers use Linux, and it's not because of Windows' licensing fees) The irony is that Microsoft itself now depends heavily on Linux. More than two-thirds of Azure customer cores run Linux, Microsoft maintains its own Azure Linux distribution, and even platforms supporting Microsoft 365, GitHub, and ChatGPT sit on Linux foundations. The same lesson applies to AI. Trying to suppress open-source/open-weight models through broad lawsuits or regulation would be like trying to ban the internet. You would not stop their development. You would merely isolate yourself, drive researchers, talent, capital, and innovation elsewhere, and become increasingly dependent on a few closed providers. Of course, genuine copyright, licensing, security, or liability violations should be addressed—but narrowly and individually. They should never become an excuse to attack open-source AI as a category. Any company or country that tries to stop open source may temporarily obstruct its own participation, but it will not stop the global movement. In the end, it will either adapt—as Microsoft eventually did—or become irrelevant. Bittensor is one of the earliest credible movers in a category that will define the next decade: open decentralized AI. Open source made the internet possible. Decentralized incentives may now do the same for intelligence. $TAO—or never.

• 
  Koya Lokendar Reddy (@Lokendar_Koya) reported 15 minutes ago

  entry-level hiring in India just hit its lowest point in years — and if you're a 2025 or 2026 fresher, you're not imagining the silence after you submit applications. here's what the data actually says, and what you can do about it. the numbers are brutal, but honest. a 2025 EY analysis found that entry-level IT roles in India have already declined by 20–25% due to automation. at the same time, a Harvard study analyzing 66 million workers found that entry-level job postings for roles requiring less than one year of experience dropped 50% between 2019 and 2024. globally, even hiring at big tech companies for fresh graduates fell by more than 50% over just three years, according to VC firm SignalFire. the WEF's Future of Jobs Report 2025 adds that 40% of employers expect to reduce staff in areas where AI can automate tasks. this isn't a blip — it's structural. India's campus placement season is feeling it hard. recruitment by prominent companies dropped by more than 50% in the 2025 season, leaving students at even well-regarded colleges sitting with uncertainty. private engineering colleges saw placement declines of 50–70% after major IT firms scaled back fresher intake, according to an Economic Times analysis. and at Infosys — one of India's biggest fresher employers — employees aged 30 and below now make up just 50.7% of the workforce, the lowest proportion in 15 years, per a Mint analysis of annual reports. until FY18, that number was consistently above two-thirds. the reason is uncomfortable but makes complete sense. generative AI is disproportionately good at exactly what freshers used to be hired to do — routine coding, software testing, basic documentation, data entry, content moderation. Harvard economists call it "seniority-biased technological change" — AI is eating the bottom of the career ladder while senior employment at the same firms keeps growing. the learning curve that used to happen on the job is now being automated before a fresher even walks through the door. but here's the part most people miss — and it matters enormously. the overall intent to hire freshers in India is still at 73% for HY1 2026, per the TeamLease EdTech Career Outlook Report. foundit's tracker shows AI-linked hiring is projected to grow 32% year-on-year in 2026 to nearly 3.8 lakh roles. NASSCOM data shows fresher hiring in AI/ML specifically grew 22% year-on-year. the demand gap is real — demand for AI engineers is rising 40% year-on-year while the skilled talent pool grows at only 15–20%, according to Taggd's 2026 salary analysis. that mismatch is your window. the jobs aren't gone. they've moved upstairs — and you need to follow them there. so what should a fresher actually do right now? five things, in order of impact: 1. build a proof-of-work portfolio, not a certificate wall. the TeamLease EdTech HY1 2026 report says hiring has shifted from "degree and resume filters" to "skills, proof-of-work and behaviour." project-based hiring is up 38% over the past year per the India Skills Report 2026. a Tier-3 fresher with three production-ready GitHub projects will beat a Tier-1 grad with a blank resume. this is no longer a hot take — it's how screening actually works. 2. get AI fluency, not AI panic. employers now specifically prioritize AI fluency, cloud & DevOps capability, cybersecurity awareness, and data intelligence as fresher hiring criteria, per TeamLease EdTech. for AI/ML roles, freshers with Python, real projects, and hands-on GenAI experience are landing ₹6–12 LPA offers, with strong portfolios at product companies going up to ₹15 LPA. 3. stop relying on campus placement as your only path. off-campus hiring is how most product roles actually get filled. 70% of off-campus roles at product startups are filled via internal referrals before the job even gets indexed on Google, per analysis of the Indian hiring ecosystem. your LinkedIn, your GitHub, your presence in developer communities — these are the actual funnels. 4. fix your resume for ATS before anything else. most Indian freshers' resumes aren't being parsed correctly by systems like Workday or iCIMS used by Amazon India and Accenture. if your resume doesn't match at least 80% of the JD keywords, a human recruiter may never see it. this is a fixable problem that costs you nothing but 2 hours of effort. 5. pick a domain + AI combination. domain expertise in healthcare, finance, or logistics combined with AI skills is more valuable than pure CS backgrounds for many specialized roles, per OdinSchool's 2025 hiring report. if you're a commerce grad, learn AI in finance. if you're in life sciences, learn AI in healthcare. the generalist AI fresher is competing with everyone. the domain-specific AI fresher is competing with almost no one. the honest reality: the market isn't punishing freshers for being freshers. it's punishing freshers for being interchangeable. the old model — join a campus drive, get a mass-hire offer, learn on the job — is dying. the new model rewards people who show up having already built something real. the window to get ahead of this is 6–12 months of focused skilling. after that, the cohort of people who figured this out gets much bigger and harder to differentiate from. if you're a fresher reading this: what's your current plan — wait for placements to recover, or go build something right now? 🎯

• 
  Snax (@aRobotNamedSnax) reported 19 minutes ago

  Hopefully helpful for someone. TLDR for most. I’ve been refining this process over the last few months because I use both Codex and Claude Code on the same VPS. I also use Cowork and ChatGPT Work heavily, often switching between a laptop, desktop and direct work on my VPS. I wanted both apps to understand what had already been done on CC or C, contribute useful updates and pick up where another session stopped. This all without maintaining separate “memories” that eventually contradict each other. The system now works around one shared knowledge base. Each computer keeps a synchronized local copy. Cowork and ChatGPT Work can both read the same project facts, operating rules, decisions, hypotheses and recent VPS activity. Cowork keeps its own product skills and uses a small autolog skill for the shared workflow. ChatGPT Work reads the instructions stored with the knowledge project automatically whenever I start a session inside it. When either app performs meaningful work, its instructions require it to create one raw session note. The filename identifies the app, computer, timestamp and topic. Meaning a Cowork session on my laptop can’t overwrite a ChatGPT Work session on my desktop. The note is updated with concise checkpoints after real changes, decisions or blockers, followed by the final outcome. This is intentionally not a transcript recorder. Normal conversation isn’t uploaded. The apps record durable work that another session may actually need. A Windows task runs every 15 minutes on each configured computer. It scans for new or changed raw notes from either Cowork or ChatGPT Work and sends them through an authenticated endpoint to my VPS. Both apps use the same uploader; there isn’t a separate synchronization system for each one. The VPS validates the project, filename, file type and size before accepting anything. It only permits writes into raw knowledge areas. Local apps cannot use this route to rewrite canonical project knowledge, change rules, upload their private skills or write somewhere unexpected. Once accepted, the VPS stores the note under the correct project, records a receipt in the shared activity history and commits the change. If an upload fails, the local file remains in place and the next scheduled run can try again. If nothing changed, the task simply reports that there is nothing to ship. The VPS is the source of truth. GitHub is the verified distribution mirror, not a competing authority. After the VPS commits an update, it publishes a verified snapshot to GitHub. The laptop and desktop periodically pull that mirror, which gives Cowork and ChatGPT Work the same updated context on both computers. The pull process also uses a ship-first rule: any unsent local raw work is uploaded before the local mirror refreshes. It refuses to silently reset over unsent tracked changes, protects raw files during cleanup and keeps recoverable collision backups when a local file and incoming GitHub file overlap. At night, Opus reviews only raw files that are new or have actually changed. Confirmed, durable information is incorporated into organized project knowledge. Uncertain findings are placed into hypotheses rather than being presented as facts. The compiler tracks file hashes, so unchanged material is not repeatedly sent back through a model. If nothing changed, it makes zero model calls. Work performed directly on the VPS follows a shorter route. That agent updates the canonical system and records the result directly in the shared changelog. It does not create a second local raw note for the same work. The next verified GitHub mirror brings that VPS activity back to Cowork and ChatGPT Work on both computers. So the complete loop is: Cowork or ChatGPT Work creates a unique raw note on the laptop or desktop → the shared Windows task uploads it within 15 minutes → the VPS validates, stores and records it → the VPS publishes a verified GitHub mirror → the laptop and desktop receive the update → Opus organizes new material overnight → future Cowork and ChatGPT Work sessions read the improved shared knowledge. The important part is what I didn’t add: no second memory database, no separate uploader for ChatGPT, no service constantly prompting models, no automatic transcript archive and no loop repeatedly processing the same information. It’s one VPS-controlled knowledge system, one shared PC uploader, one GitHub mirror and one nightly changed-files-only organization pass. Cowork and ChatGPT Work keep their own strengths, but they now work from and contribute back to the same history. Hopefully this helps someone dealing with the same problem

• 
  Radial (@radialbuild) reported 20 minutes ago

  If GitHub Issues still fits, stay. Free, already there, fast, right next to the code. The day a flat list stops scaling is the day you start losing things in it. You do not have to leave the repo to fix that. Radial links to your branches, PRs, and commits.

• 
  Enes (@enesozturkdev) reported 20 minutes ago

  Plan for today; ship like crazy for side project Meanwhile two pillars of it GitHub and OpenAI down so bad

• 
  Mikhail Rogov (@i_mika_el) reported 23 minutes ago

  @Felirami @steipete @openclaw Add a GitHub Sponsors or Buy Me a Coffee link beside Arca's OpenClaw issue history, so people who see the work can support you directly.

• 
  Glenski 📷🇨🇦 (@glenegrant) reported 30 minutes ago

  @araseb_ I actually have both working together: Codex on macOS, Claude Code on Windows PC they work through GitHub issues on a /loop as we port applicaitons from Win to macOS. Kind od wild to witness.

• 
  Teddy - PolyBackTest.com (@Bitcoin_Teddy) reported 31 minutes ago

  A guy named nbatman on Reddit accidentally built the most useful website on the internet. It's called FMHY (Free Media Heck Yeah). This is the website Google delisted from search for DMCA violations, Reddit shadow-banned for promoting piracy, the Motion Picture Association flagged as a top piracy threat, and the RIAA pressured hosting providers to drop. It is still online. It is still updated every month. Here's how it works. FMHY is the index. The wiki itself hosts nothing. It just tells you where every free thing on the internet actually lives, organized into 14 categories with safety ratings on every single link. → Movies and shows in 4K from 50+ streaming sites → Music at Spotify and Apple Music quality → Adobe Creative Cloud, Microsoft Office, AutoCAD, JetBrains → Every paid course on every major learning platform → 100 million books and papers through Anna's Archive → Free alternatives to every paid AI tool → A SafeGuard browser extension that flags unsafe sites in real time It started as a single Google Doc maintained by one Reddit moderator in 2018. Google killed it with a DMCA takedown in 2023. The community rebuilt the wiki on its own domain, mirrored it to GitHub and IPFS, and now runs it across 12 backup domains simultaneously. There is no company. No CEO. No central server. Six anonymous volunteers maintain the entire thing in their spare time. Donations through Ko-fi pay for the hosting. Nobody profits. Hollywood can't shut this down. Spotify can't shut this down. Adobe can't shut this down. The entire subscription economy is held together by you not knowing this wiki exists.

• 
  Jeff Schneider (@jeffrschneider) reported 32 minutes ago

  @doodlestein @FreeDrThug I think @charlespacker had to solve the problem in a specific way, and built it. Searching github for a look-alike solution, and evaluating them all is a non-starter. You know this better than anyone.

• 
  Samraaj Bath ⚡️ (@samraaj) reported 35 minutes ago

  @nedoleary I've realized this is often an incentive problem because someone's "*** is on the line". In your recruiting example, they are dismissive because they need the clean story for the negative case. If a no-name candidate doesn't work out, they're screwed bc they assumed the risk. And it doesn't even make sense to take that risk bc their upside is capped (static commission) and the downside is unlimited bc they get fired. If a credentialed candidate doesn't work out, well "look at their github!" or "they were a FAANG engineer!". Show me the incentive, i'll show you the outcome.