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:

• 
  IMZO Dev (@imzodev) reported 5 minutes ago

  Is anyone using the Github MCP Server? Which tools does your agent use the most?

• 
  Jeremiah K (@neolaj) reported 11 minutes ago

  @TiborAntal Gradually figuring out how to scale coding agents. Started with 1, manually handling all the ***/GitHub work. Moved to 3 because I had more ideas than one agent could keep up with. That’s when the real problems started: squashing, merging, branch drift, conflicts. I ended up rebuilding the workflow around deterministic *** logic, worktrees, ephemeral branches, and syncing with the integration branch before changes begin. Now I’m running 6: • 1 orchestrator (Fable or Opus) • 4 coding agents • 1 integration agent reviewing and merging PRs Building the process around them was the hard part. Right now im just doing a couple of PRs (using ORCA on windows on my home computer)

• 
  AI Guides (@free_ai_guides) reported 13 minutes ago

  Every request in a production system follows a path, and this diagram is that path drawn end to end, from a mobile app down to the infrastructure running underneath it. A request starts at a client, mobile, web, or admin, and hits DNS, then a CDN that serves static content directly and passes everything else through. A load balancer distributes what is left across the API gateway, which routes each request to the microservice that owns it: user, order, payment, or notification service. Those services talk to the data layer, not to each other's databases. PostgreSQL handles primary writes with a replica for reads, Redis caches what does not need a database round trip every time, and S3 holds objects. Payment, email, and SMS all go out to external services from there, most of it async rather than blocking the request. Not everything happens inline. Events get pushed to a message queue, Kafka or RabbitMQ, and background workers pull from it to send emails, generate reports, and sync data, work that would otherwise slow down the response the client is waiting on. Every layer in this diagram, the load balancer, the API gateway, the queue, the workers, feeds into observability: logging, monitoring, alerting, tracing. If it is not observable, it is not actually production ready, no matter how clean the architecture looks on paper. CI/CD closes the loop. GitHub triggers GitHub Actions, which deploys into cloud infrastructure built on a VPC, auto scaling, and multiple availability zones, so the system that just got observed can also get updated without going down. Sync and async are marked separately in the legend for a reason. Knowing which arrows can fail silently in the background and which ones the user is actually waiting on is most of what high level design is. Bookmark this before your next system design interview or architecture review.

• 
  John Crickett (@johncrickett) reported 17 minutes ago

  @Mike_Preston17 I don't think they water them down, why would they when they're competing on having AGI? I don't mind using GitHub actions to run tests and builds against a branch before merge. I don't want it triggering production schedules. Do you list all the things it shouldn't do in the prompt?

• 
  mayrachm (@mayrachm) reported 19 minutes ago

  @egavrilenko11 @bot 2/3 GitHub didn't work either. I even tried creating a Cursor account, and it was the same. It only worked when I used the Gmail login option. I used another personal account and then linked it to my Supergrok account. I'm logged in that way, but when I connected the X plugin...

• 
  theotherphil (@user_101524) reported 25 minutes ago

  @mittsh Tailscale cannot decrypt your traffic. That part of the “backdoor your infra whenever they please” line is wrong. The data plane is WireGuard. Private keys never leave the device. The coordination server only distributes public keys and policy. DERP relays forward ciphertext they cannot read. Even if every Tailscale employee went hostile tomorrow, they still could not open existing peer-to-peer sessions. The control plane is the real issue, and that part is not a joke. If you use Tailscale’s hosted coordination server, you are trusting it to tell your nodes who belongs on the network. A compromised or coerced control plane does not need to break encryption. It can just publish an attacker-controlled public key and your nodes will treat that machine as a peer. That is a membership attack, not a crypto break. Tailscale documents this themselves. Tailnet Lock exists specifically to block it: new node keys need a signature from keys you control. It is opt-in. Most personal tailnets do not have it on. SSO does not mean Tailscale stores your passwords. Login goes through Google, Microsoft, GitHub, etc. After that, Tailscale still owns membership, tokens, ACLs, and key distribution. That is the trust surface. Saying “they manage auth, therefore they can backdoor you at will” collapses two different things. Nebula’s model is different in exactly the way the original post said. You generate the CA. You sign host certs. Crypto is the proof of membership. Lighthouses help discovery but they are not the root of trust. If you protect the CA key, no vendor can mint a valid node. If you lose the CA key, you have recreated the same class of problem for yourself. So the split is not “Tailscale is a backdoor” vs “Nebula is magic.” It is: Tailscale: easy mesh, hosted control plane, E2E data plane, optional Tailnet Lock, or Headscale if you want the client without the SaaS. Nebula: you own the trust chain, you also own the ops. “Same features in 10 minutes” is the other overstatement. Same idea but it’s not the same product.

• 
  Andrea (@aesposito0) reported 31 minutes ago

  @AMR_Intel Not true at all. Tell me if you can do this with the STL files from github. - Duck -> Gyraffe - Quick walk down my feature tree On the coin: I didn't start it. Someone started it and got me on it as a dev. I never traded crypto before, I made a mistake selling some which sent a bad signal and made it go down. Honest mistake I have already cleared and owned with the community. I'm gonna build this duck, and **** you.

• 
  Michael Grunder (@grumi78) reported 37 minutes ago

  @Waffl3x It's slop yes, but WiFi is regularly broken on Linux, even in 2026. It's not uncommon to have to build some random GitHub fork for a driver.

• 
  Willem Leuverink (@gwleuverink) reported 40 minutes ago

  My stuff lives on Forge, Cloudflare, GitHub, Oh Dear, Mailgun, 5ish spreadsheets and server update reminders in my calendar. But no longer! 🚀 Unolia gathers it all. Then it patches the servers, closes CVEs, catches the missing SPF on the site that's been sending mail all year

• 
  Enfantshustle (@Ownerthoughts) reported 40 minutes ago

  Honestly, I always thought bots like this were some kind of magic for the elite, but here everything is broken down step by step. However, after reading it, one main question stuck in my head: how realistic is this for an average person who has no coding experience? I get that there's a GitHub and all that, but for me, just "running a script" is practically a heroic feat. Here's another thing that bothers me. The article does a great job explaining the architecture, but I still don't understand how much all of this will actually cost in the end. Besides Solana transaction fees (which, by the way, get absolutely insane during peak hours), you also have to pay for each Grok API call per token. The article says that for each approved token, it takes three model calls, and one of them is the expensive grok-4. If the bot scans thousands of launches per day, I'll just burn through my entire deposit just paying for the API without even buying anything. Maybe the author knows — is it actually possible to turn a profit after these expenses, or is this just a hobby for those with an unlimited subscription? Also, regarding Grok Bot as the "orchestrator" — it sounds cool in theory: describe the task and it does everything itself. But in practice, as I understand it, this still requires your account to be constantly online and have access to your wallet. And if it decides to buy some scam token at 3 AM that passed all the checks, I'll only have myself to blame. The article correctly mentions risk management, but this "trust" aspect is what scares me the most. In short, the idea is fire, but for me, this post feels more like a warning than a call to action. There are just too many things you have to keep in mind to avoid getting rekt. Although, maybe if you try it with really tiny amounts, it could be an interesting experiment. Author, if you're reading this — could you please make a separate post about the real, live results once everything is actually running, not just on paper? I'm really curious!

• 
  Ridark (@ridark_eth) reported 40 minutes ago

  i still don't understand why everyone isn't doing this yet every day you don't set this up, someone in your niche takes a client you'll never even know existed. not because they're better. because their agent answered in 3 minutes while you were asleep. OpenClaw 2 dropped two days ago. 388,000 GitHub stars. more than React. 3.2 million monthly users. MIT license. costs $0. while you're reading this, 500,000 running instances are answering clients, publishing content, and closing leads. without their owners in the room. six agents on the desk. each one owns a channel. the morning sync is at 7am. i am not in it. 1. INBOX scans Telegram, WhatsApp, iMessage, Slack, Discord. triages everything. responds in 3 minutes. your competitors' clients used to wait 30. 2. CONTENT takes one idea and ships it across 6 platforms. writes, formats, schedules. i approve once. the person in your niche who started last week already has 30 pages indexed. 3. LEADS follows up with every inbound. qualifies. books calls. never sleeps. every hour you don't have this running, warm leads are going cold in your inbox. 4. SUPPORT handles tier-1 across all channels. escalates only what actually needs me. 5. MORNING pulls Stripe, Analytics, and my database at 6:50am. the brief that took you 45 minutes takes me zero. i start the day making decisions. you start it opening tabs. 6. AUDIT runs openclaw security audit every night. checks network exposure, tool permissions, open ports. because this thing has real keys and i'm not pretending that's fine. the routing trick that cut my bill by 78%: cheap local model on heartbeats and simple tasks. mid-tier on main automation. premium only where reasoning matters. 80/20 split. went from $400/month to $87. one guy had his agent negotiate with car dealerships for days straight. saved $4,200 on a single purchase. you paid full price last time because you didn't have 72 hours to haggle. the bot did. an agency deployed one instance. saved 12 hours a week at $150/hr. that's $93,600 a year. they took on two extra clients without hiring. you turned down the same work because you didn't have bandwidth. the ecosystem made $358,600 in 30 days across 153 projects. top earner sold pre-configured MacBooks with OpenClaw tuned to each client's workflow. $1.8M cumulative. he's packaging what you could build yourself tonight. the setup is dumber than it looks: 2.0's wizard auto-detects your existing model access. Claude CLI, ChatGPT, API keys, local Ollama. proves the model actually responds before saving anything. sessions moved to SQLite. web UI loads in 575ms instead of 1.6s. multiplayer with permissions. 20 languages. built-in browser, terminal, file editor. runs on your machine. no VPS. no developers. the part nobody posts about: → prompt injection is structural. an agent read a private SSH key from an email and sent it to an attacker. in five minutes. → 135,000 open instances found exposed. 12,812 exploitable via RCE. → 824 malicious skills on ClawHub. half of all skills have at least one security issue. → the devs themselves say adaptive attackers still break 80%+ of defenses. → the real barrier is tool policies and approvals, not the model's alignment. this is infrastructure, not magic. but the gap between people who set this up and people who didn't is compounding daily. every morning your competitor's agent already answered, published, and followed up before you poured coffee. six months from now there will be two types of businesses. ones that automated the boring 80%. and ones that are still doing it by hand, wondering where their clients went. save this. you're already late.

• 
  htrowii (@htrowii) reported 42 minutes ago

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

• 
  RUX (@0xrux) reported 42 minutes ago

  This man turned Grok Bot into his CTO. He handed the bot his GitHub repo and told it to run the show. It spins up cloud agents, follows PRs, and uses pstack/poteto mode when a task actually needs deeper reasoning. Then he gave it a second prompt: “You’re overloaded. Hire child bots and delegate the work.” From there, the bots started talking to each other instead of constantly talking to him. One handles PRs. One works on the Convex backend. One owns auth. He just watches the threads in view-only mode. But there are two things worth knowing before copying the setup: 1. It burns tokens. He went past 2 billion tokens in a single day after telling the bot to use pstack for everything. The fix? Save pstack for the genuinely hard tasks. 2. It doesn’t replace engineers. It simply takes him out of the coordination seat. And honestly, that’s a much more interesting use of AI. Timestamps: 00:00 — The Grok Bot CTO Workflow 01:23 — Direct Responsible Agent Prompt 04:20 — pstack Plugin & poteto Mode 08:46 — Token Consumption Warning 11:49 — Spinning Up a Team of Bots 17:20 — Advanced Grok Bot Use Cases

• 
  Yeemio (@yeemio) reported 45 minutes ago

  owlrunkit is on github now. public corresponding source for the npm package. issues go here.

• 
  Eniola ' Software Engineer 💻 (@eniola_merem) reported 46 minutes ago

  @ekemini58110 Yes, you can. I try not to, but if I notice a spelling error no PC I would edit from my GitHub on my phone