GitHub status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of GitHub reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Most Reported Problems

The following are the most recent problems reported by GitHub users through our website.

• Website Down (62%)
• Errors (21%)
• Sign in (18%)

Live Outage Map

The most recent GitHub outage reports came from the following cities:

City	Problem Type	Report Time
Tlalpan	Sign in	3 days ago
Quilmes	Website Down	3 days ago
Bengaluru	Website Down	5 days ago
Yokohama	Sign in	6 days ago
Gustavo Adolfo Madero	Website Down	10 days ago
Nice	Website Down	10 days ago

Full 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:

• 
  Mi.lu. (@ludwim_i) reported 2 minutes ago

  Sorry guys, here is a quick status update. I planned to release a bigger update for the Robot Skill Registry today, including the GitHub and Hugging Face integration. The idea is that you can connect your GitHub and Hugging Face accounts with the app. This should make it easier to search for things related to your robot setup, such as repositories, data models, policies, and other relevant resources. Unfortunately, the integration is not working reliably yet, so I need to do some more coding and testing. Because of that, I won’t release it today as planned. I’m sorry for the delay. Maybe I can release it after the weekend, but I don’t want to push something that is not ready yet. If anyone has feedback on whether this direction makes sense, I would really appreciate it.

• 
  Chandraprakash Darji (@chandra_7852) reported 5 minutes ago

  @raycast Feature Request: Bangs, like in alfred we can hit space and search something it will list only file. Something like thies we can set bangs for file search, clipboard, github issue serach. It is hard to set the shortcut for everything

• 
  Kunal Yeola (@yeolakunal) reported 5 minutes ago

  Asked GitHub Copilot to fix ESLint issues and it added eslint-disable at the beginning of the file 😭

• 
  John Evans Okyere | TheAISolutionist (@Ananselab) reported 7 minutes ago

  Deployment failed with: dial tcp :22: i/o timeout The app was fine. SSH was fine. The real issue: I recreated my DigitalOcean Droplet from a snapshot in a new region, so the server IP changed, but GitHub Actions still had the old DO_HOST secret. Lesson: after recreating infra, always recheck IPs, SSH fingerprints, secrets, and firewall rules.

• 
  Kamil Skowron (@kamilskowron) reported 8 minutes ago

  @donutkiller_pro If you could - always confirm with the current main branch. If the issues show up on that branch - it would be best if you could dump the generated report into GitHub issues please 🙏🏻

• 
  Moe Sbaiti (@MoeSbaiti) reported 9 minutes ago

  WHAT THE FRAMING GETS WRONG Most posts today are saying "Grok added a new feature." That framing is backwards. What happened is that an agent framework with over 110,000 GitHub stars, the number 1 ranking on OpenRouter, and an NVIDIA endorsement just got native access to one of the most capable models available through a simple OAuth login. xAI made the announcement. Not Nous Research. Hermes Agent also self-improves. When it solves a hard problem, it writes a skill file for that solution and saves it. The longer it runs on your specific workflows, the more capable it becomes for your specific context. That is not how people are talking about this today. The memory layer and the self-improvement loop are the actual product. Grok is the engine.

• 
  Jerome (@jeromeq2004) reported 10 minutes ago

  github releasing the agentic ai developer cert is funny because the actual exam is going to be 'fix this thing claude broke in production while it tells you the tests pass'

• 
  Dewaldt Huysamen (@GodsBoy7777) reported 11 minutes ago

  @sickdotdev Getting insane and better results just on medium for all of the above categories. Weirdest is Opus 4.7 fails at basic school tasks help for kids and when I do code GPT 5.5 finds issues that are found in any case on github CI checks. If use codex CI passes more than 99%

• 
  Nicholas Losciuto - Dog Dad (@thedogfather) reported 13 minutes ago

  @Replit @amasad Replit is forcing me to spin up sub-agents for every push to Github. First, cost $10 just to rebase. Then 6 failed commits for two edits that were 5/10 complicated. Each requiring a new sub-agent to push to ***. I'm also out an hour of my time. Same issues on my other apps. Already had to switch to Codex but came back to try again, seems worse. Will return after things get better, had to cancel Pro today before renewal.

• 
  Justin Cromer (@just_cromer) reported 14 minutes ago

  @htmx_org github is down sry

• 
  Dusko Licanin (@dulelicanin) reported 15 minutes ago

  Your AI wastes 65% of its tokens saying "Sure, I'd be happy to help you with that." A 19-year-old developer made a markdown file that fixes this. It got 12,000 GitHub stars in 4 days. Here's what happened: Julius Brussee created "Caveman" — a Claude Code skill that forces AI to talk like a caveman. No articles. No filler. No pleasantries. Just the technical answer. Before (69 tokens): "The reason your React component is re-rendering is likely because you're creating a new object reference on each render cycle. I'd recommend using useMemo to memoize the object." After (19 tokens): "New object ref each render. Inline object prop = new ref = re-render. Wrap in useMemo." Same fix. 72% fewer tokens. But here's what most people miss: A March 2026 paper (arXiv:2604.00025) tested 31 LLMs across 1,485 problems and found something wild: Forcing large models to be brief improved accuracy by 26 percentage points. Bigger models literally perform WORSE because they over-elaborate. The researchers call it "scale-dependent verbosity" — the model rambles, and rambling introduces errors. Less words = more correct. Not a meme. Peer-reviewed science. The real cost math: → Anthropic charges 5x more for output tokens than input → 10,000 API calls/day at 150 tokens each = $8,212/year → With caveman compression = $2,847/year → Savings: $5,365/year per agent And here's the honest part most viral posts won't tell you: The 75% reduction only applies to isolated chat responses. In real coding sessions, independent benchmarks show 14-21% savings on output, and ~25% on total session tokens. Still meaningful. Still worth it. But not the headline number. The deeper insight? Chinese developers have had this advantage all along. Chinese has no articles, no verb conjugation, and each character carries more semantic weight. Chinese prompts naturally use 30-40% fewer tokens than English. Caveman mode is essentially porting the token-efficiency of Chinese into English. We spent billions training AI to be eloquent and polite. Now we're paying $15/million tokens for that politeness. The most sophisticated AI systems ever built — made to grunt through code reviews. That's the real story. Link to the repo and the research paper in comments. What's your take — does forcing brevity help or hurt AI reasoning?

• 
  Lu (@LuminousTheReal) reported 16 minutes ago

  @SilverYogensha @thsottiaux Potentially, but it seems like it's using 5.4 to compact it since 5.5 is bugged out. I found it on github, lots of people are suffering with this compacting issue and i think we will get a reset soon since tibo mentioned the quality si down

• 
  Abu Olumi 🪶 (@Olumi441) reported 17 minutes ago

  There's also a public feed. BaseLens fetches Base GitHub releases and analyzes them with AI automatically. Clean upgrade cards. No jargon. No noise. Anyone can read it, no login needed.

• 
  Yaseen Shaik (@YaseenTech4) reported 18 minutes ago

  Just completed an assignment on building a dependency graph for AI agent tools using Google Super + GitHub integrations 🚀 Started with: “This should be easy” Then came: TypeScript errors zip/upload issues CRLF debugging 😭 finally got the submission accepted successfully ✅

• 
  Roj Niyogi (@niyogi) reported 20 minutes ago

  @colinhacks @pullfrogai @Pullfrog so for 90% of folks who are using cursor/windsurf/github copilot/claude code via an IDE "chatting" with their agent, the answer is just to tell your agent to fix everything? i've used both coderabbit and greptile extensively and here's what happens: 1. 20% of what is found is false (and likely even more variable if you pick a model) 2. i cherry pick issues and paste in editor to handle 3. code review is rarely "happy" and you can end up in a loop state that burns tokens bummer that you've got the flexibility on the one end but have a strong opinion on the other for what seems to be an obvious opportunity to close the loop for, i'd guess, a chunky subset of users

• 
  Java (@rishabhjava) reported 24 minutes ago

  @github How about the existing product stops going down first

• 
  Sunil Thakare 🇮🇳 🦀 (@thakares) reported 24 minutes ago

  @moneycontrolcom @Copilot Microsoft's AI ecosystem, centered on Copilot (across Windows, Microsoft 365, GitHub, Edge, etc.), has faced significant criticism for underdelivering on hype, poor adoption, reliability issues, security/privacy problems, and high costs relative to value.

• 
  J Filipe (@jrmromao) reported 26 minutes ago

  Pivoted CostLens from "AI cost tracking" to "AI productivity measurement" last week. Built in 5 days: - MCP server that tracks what AI agents actually ship - Automated ROI reports for engineering leaders - CLI setup in 30 seconds - GitHub PR correlation Same product, completely different value prop. Before: "save money on AI" Now: "prove AI delivers value" One resonates with finance. The other resonates with everyone. #buildinpublic

• 
  Neo (@neodevils_) reported 27 minutes ago

  @codeblue87 @diabrowser Hey, A few weeks ago, I tried to refresh to view all of my PRs in tabs with new GitHub Pull Request preview. But it was not working. I know this is a beta feature in GitHub and they might publish it sooner. Will Dia work on that before it is late?

• 
  Shubh Jain (@shubh19) reported 32 minutes ago

  @devXritesh now it’s mostly docs, blogs, github issues, and AI explanations instead of full books

• 
  Ashish Ranjan (@Ashish_050488) reported 37 minutes ago

  build on laptop (3 secs), upload only the dist folder. 500kb. server just serves files now, doesn’t build anything. deploy went from 15 mins to 5 secs. turns out big companies do this exact thing, just automated. github actions next so i never think about

• 
  Dylan from 2045 (@dylan2045ad) reported 38 minutes ago

  AI ate tech this week: OpenAI is Coke. Anthropic is Pepsi. Grok is RC Cola. Gemini is the Costco brand your dad swears tastes the same, lets see what tomorrow brings? Mistral is the European one with real sugar. Llama is the 6 pack at the back of the fridge nobody opened. GitHub is buckling under AI-generated PRs. Hashimoto is out. Polymarket whales hide behind VPNs and crypto, but AI agents are chasing insider traders and hunting them down. We built the thing that's eating us.

• 
  Kristopher Betz (@kjbetz) reported 42 minutes ago

  @davidfowl I think I do... I push code to GitHub. Actions kick off, build new containers, build new migraines, then self hosted runners pick it up and run migrations, and auto update containers which pull down new images and restart containers.

• 
  Jeff Hayes (@JD__Hayes) reported 45 minutes ago

  @FredKSchott I'm interested, but web page is down and could not find on github.

• 
  𝒹ℯ𝓁𝓁𝓎_𝓉𝒽ℯ_𝒹ℯ𝓈𝒾𝑔𝓃ℯ𝓇 (@dellyricch2) reported 45 minutes ago

  Elon says the latest 𝕏 algorithm has been published to GitHub Can someone please break it down for us

• 
  Your Friend Myk (@mykola) reported 46 minutes ago

  @joelhooks is this just static content? so like a github pages alterntaive? can't run a server etc?

• 
  The Doctor (@Doctorthe113) reported 46 minutes ago

  @_yorunoken @sebastienlorber On my chromium and Firefox based browsers I only see a slight delay/flickers. Probably caused by the huge number of texts. I don't see how GitHub can fix that without using something like pretext. But what op posted seems to only affect macos. Unfortunately macos users tend to be entitled people. So people like op will continue to blame GitHub instead of using a better browser or complain to chromium.

• 
  David Cramer (@zeeg) reported 50 minutes ago

  @eternalmagi dont have context, is there a github issue by chance

• 
  AtomicNodes (@AtomicNodes) reported 52 minutes ago

  Hermes Agent vs OpenClaw on Qwen 3.6 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb. OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw: hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes: parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

• 
  Hiren Thakkar (@thehirenthakkar) reported 57 minutes ago

  Microsoft's GitHub got a 400% increase in organic traffic. No new content. No link building. No redesign. They fixed cannibalization. Blog posts competing with marketing pages. Multiple pages fighting for the same keyword. None winning. Removed the mess. 400% increase. Check it today for free: → Screaming Frog: crawl your site, export titles + H1s + canonicals. Two pages targeting same keyword? One is killing the other. Missing canonicals? Even worse. → Detailed SEO Extension: free Chrome plugin. One click shows canonical, title, H1, H2s. → Google Search Console: filter by keyword → Pages tab. Multiple URLs? That's cannibalization. Fix: pick strongest page. Redirect or no-index the rest. Add missing canonicals. Sometimes you don't need more content. You need the basics done right. Try it this weekend and tell me what you find. (GitHub case study via Brain Labs.)