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:

• 
  LeanCTX (@leanctx) reported 1 minute ago

  Crossed 860 GitHub stars on lean-ctx. Wild that something I built to fix my own token bill is now used by thousands of devs.

• 
  Bean Juice Studios (@BeanJuiceStudio) reported 3 minutes ago

  @Alexzoin Yeah two sources is usually perfectly fine. The odds of your GitHub account getting hacked or the platform going down is extremely very low lol

• 
  Amu (@amu4biz) reported 4 minutes ago

  Github down again! grrrr using @gitlawb asap

• 
  Upwind Security MDR (@UpwindMDR) reported 6 minutes ago

  🚨 Critical SSRF in GitHub Enterprise Server (CVE-2026-9312) A server-side request forgery (SSRF) vulnerability in GitHub Enterprise Server allows an unauthenticated attacker to send crafted requests to internal services by exploiting insufficient input validation in an upload endpoint. Attackers can use path traversal to bypass restrictions and redirect internal API calls, potentially accessing sensitive internal services and exposing credentials. 👉Affected: GitHub Enterprise Server < 3.16.20, < 3.17.17, < 3.18.11, < 3.19.8, < 3.20.4, < 3.21.1

• 
  Beau Johnson (@BeauJohnson89) reported 6 minutes ago

  coding agents are getting repo onboarding kits now baskduf/harness-starter-kit > 32 stars on github > created yesterday > mit + python > repo-native harness engineering starter kit > turns repeated chat instructions into durable repo artifacts > adds agent rules, drift checks, knowledge storage, feedback loops, and adoption reports > supports python, typescript, nextjs, django, fastapi, spring, react, vue, and generic repos the useful part is the framing: most teams keep trying to fix bad agent work with better prompts this repo treats the repo itself like the product surface for agents rules live in files checks run locally mistakes get remembered updates have a workflow the next serious coding agent setup wont be just pick claude or codex itll be model + harness + memory + verification thats the stack

• 
  Adam Peterson (@_adam_here_) reported 7 minutes ago

  And... @github down. @githubstatus I can't push.

• 
  Serena Ge (@serenaa_ge) reported 8 minutes ago

  We wanted tasks that reflect realistic, novel engineering work. The SWE-Bench family scrapes existing GitHub issues and PRs, which creates two problems: memorization (models have already seen the solution) and triviality (most tasks are small). DeepSWE tasks are built from scratch, keeping prompts intentionally short and natural while requiring significantly more code to solve.

• 
  João Capinha (@joaopcapinha) reported 8 minutes ago

  I'm using Claude Code more than any other tool in my stack right now, and not just for writing code. Here's how I'm running a DeFi project with agentic AI orchestration at its core. I'm coordinating across ClickUp, Google Drive, Slack, GitHub, and call transcripts. And the problem isn't that the information doesn't exist. It's that finding it, connecting it, and moving it between tools quietly becomes the job. Nobody puts "context archaeology" in the job description. It just eats your day. The shift: instead of jumping between apps, I talk to Claude about the project. Pull the latest task status. Cross-reference a spec. Update a ticket. Draft a message. One thread. No rebuilding context from scratch every time. It also makes you faster at being wrong (worth saying out loud!) Feed it a messy problem, you'll get back a very confident-sounding mess. The judgment still has to be yours. But the coordination overhead? The invisible tax that was never really the job? Most of it has disappeared. Huge productivity unlock.

• 
  Oliver (@fomoliver) reported 10 minutes ago

  A guy found a bug in Claude and instead of opening an issue on GitHub he went on stage. The room cracked up. On his screen Claude Code was switching between .Claude and .Claire. Back and forth. Over and over. Because Claude runs without memory, without rules, without guardrails — for 99% of people who launch it every day. It invents names. Writes files wherever it wants. Forgets what it did in the last session. Three settings that put it in line in 5 minutes: 1. CLAUDE.md in the project root A text file with rules for Claude Code: "don't touch /legacy", "tests go in /tests", "TypeScript only". Claude reads it before every task and follows. Fixes invented names and file chaos. 2. Memory MCP server Gives Claude memory between sessions — it remembers yesterday's decisions, your code style, bugs already fixed. Installs with one command. Fixes "forgets what it did last session". 3. Permission rules in settings.json You spell out what Claude can and cannot do without asking: "don't run *** push", "don't touch .env". It asks before every such action. Fixes chaotic file actions. 5 minutes per setting. Claude stops being a junior on steroids and becomes a controlled tool. Bookmark this. You'll get why in a week.

• 
  Dastaan (@akhilsinghind) reported 11 minutes ago

  Github & Cloudflare. Read. Yea man, when you stayed on GitHub + Cloudflare. Cheap servers get cooked fast when bots start going crazy. Since my client's site got no login now & I patched the repo input, there ain't much left to hit besides straight spam traffic. What?

• 
  Avish Hakani (@AvishJH) reported 18 minutes ago

  Microsoft is reportedly winding down many internal Claude Code licenses after heavy internal adoption and shifting teams toward GitHub Copilot CLI. Interestingly, the issue wasn't that employees disliked it. Employees loved it.

• 
  Boxing Avocado (@rodohale) reported 19 minutes ago

  @rauchg the problem with github is that they are taking a lot of engineering time building bs nobody asked for

• 
  Theo - t3.gg (@theo) reported 22 minutes ago

  I'm going to use my AI psychosis to fix clouds for agents. Someone else needs to use their psychosis to fix source control. I would do it myself but I'm already too deep on the cloud thing. GitHub is dying and *** is not the right primitive. Will dump some thoughts here.

• 
  Darren Shepherd (@ibuildthecloud) reported 26 minutes ago

  @mjtechguy That's the worst idea. I want to leave GitHub. GitHub is just slow and painful at this point. I really don't use it anymore except for to automate my release process because I need to do releases across the Linux window in Mac, it's the easiest to do that with GitHub Actions but it's so painfully slow.

• 
  The Cyber Jim (@thecyberjim) reported 28 minutes ago

  CrowdStrike & Google just took down GlassWorm — a year-long developer-targeting botnet that poisoned 300+ GitHub repos via trojanized VS Code extensions, npm packages & Python libraries. The malware used 4 resilient C2 channels: Solana blockchain, BitTorrent DHT, Google Calendar dead drops & commercial VPS. It stole developer tokens (GitHub, NPM, OpenVSX), harvested crypto wallets, deployed credential stealers, WebSocket RATs & Chrome extensions that logged screenshots/keystrokes. Infected machines became SOCKS proxies, HVNC servers & execution nodes. All C2 channels simultaneously neutralized. Developers just dodged a bullet.