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:

• 
  Andrew (@Tradesdontlie) reported 2 minutes ago

  @dyk_ceph i might be absolutely insane but i use github issues/epics/milestones as the source of truth for planning. i do want to get into semantic flame graphs for recursive agent memory…. but we will get there lol /goal pick up issue #420 or /goal finish milestone 6.9 it has 4 parents and 12 sub issues in said issue it would list out related issues parents in the milestone essentially what modules interfaces adapters seams to work with specifically, what we are trying to accomplish. include a codebase architecture design and philosophy (The concept of deep modules and easy-to-use interfaces in John Ousterhout’s book “A Philosophy of Software Design”) helps steer all the planning The philosophy helps what not to do and the end result of the milestone via vision .md and some mermaid graphs of what of the final product design should be shows the model what happens after we do everything what’s the end result we are trying to achieve A bunch of **** really. Do this, create real testing not bullshit nonsense, follow best practices, when CI passes at the end locally submit PR Let it rip the human in the loop is only needed in certain aspects. ideally i would find a way for an agent to delegate the milestone goals to other agents, have them talk back and forth a few times until they are aligned and boom it goes off does the thing and another review agent comes in and tests the actual application frontend. graph engineering is sick just haven’t fulllllly gotten there yet

• 
  Rui Almeida (@ruibeard) reported 4 minutes ago

  @dhh @mischavdburg The Omarchy skill could include instructions to automatically submit the fix to GitHub when it actually solves drivers/HW related problems. Those could then be scanned to apply the fix upstream

• 
  József Schaffer (@angelday) reported 5 minutes ago

  @AndyHewco If you have the source on GitHub I could look into the Mac issues real quick.

• 
  grey (@w3b3grey) reported 5 minutes ago

  If you are like me and you wondering, where my IDENTITY PEM FILE is on @flop_labs , I gat you; here is how to find it identity.pem is already saved automatically; the init command writes it to your project folder the moment you run it (typically right in technocore-did-starter/identity.pem). You don't need to do anything extra for it to exist. What "saving" really means here is backing it up safely, since if this file is lost, your DID is unrecoverable (the guide's troubleshooting table says exactly that: "there is no central DID recovery service"). 1. Confirm it's there ls -la ~/technocore-did-starter/identity.pem 2. Back it up to a second location — copy it somewhere other than the working folder, e.g. an external encrypted drive or a password manager that supports file attachments (1Password, Bitwarden both do this): cp ~/technocore-did-starter/identity.pem ~/Desktop/identity-backup.pem Then move that copy off your main disk (external drive, encrypted USB, etc.) rather than leaving a second copy sitting in ~/Desktop long-term. 3. Lock down file permissions so only your user account can read it: chmod 600 ~/technocore-did-starter/identity.pem 4. What NOT to do with it Don't upload it to GitHub, Google Drive, iCloud Drive, Dropbox, or any synced/cloud folder in plaintext. Don't email it to yourself or paste it into a chat (including this one). Don't commit it to *** , the guide's Path B steps even have you run *** ls-files "*.pem" "*.key" before committing specifically to catch this. 5. Remember the passphrase separately from the file The .pem is encrypted, but it's useless without the passphrase you set during init. Store that passphrase somewhere separate from the .pem backup itself (a password manager entry, not a text file sitting next to the key), so a single leaked backup doesn't hand over both pieces at once. NOTE - If you lose either the file or the passphrase, per the guide, there's no recovery; you'd have to run init again and get a brand new DID.

• 
  Mainak Mukherjee 🇮🇳 (@MainakMukh47974) reported 8 minutes ago

  Now there is a long process that takes place between development and production. First, you write the code; then the code is uploaded to the remote server, where an executable of that code is built, and that is known as the "build" process. This is then tested and deployed. Manually doing it every time you push some new changes is very annoying and hence there is a way to automate it using CI/CD tools like GitHub Actions. So every time a new code is pushed, the build->test->deploy pipeline gets triggered automatically.

• 
  Ola (@ola_onlinee) reported 9 minutes ago

  Nvidia just bought the front door to open-source AI. The deal: $12.9B for Hugging Face, the GitHub-like repository where developers find, share, and deploy open models. Reported by The Information, two days after Business Insider said Hugging Face was exploring a sale. Why Nvidia wants it: • Defence. Anthropic and OpenAI are building their own chips to reduce dependence on Nvidia GPUs. If closed labs escape, open models become Nvidia's demand floor and now Nvidia owns where those models live. • Position in the stack. Nvidia dominates the hardware that trains and runs these models. This buys them the layer above: where devs discover models, optimise them for specific hardware, and ship them. • Default optimisation. Own the distribution hub and "runs best on Nvidia" becomes the path of least resistance for millions of developers. Is the price fair? Depends what you're buying. Against revenue, no. Hugging Face reportedly runs around $150M annualised. That's roughly 86x revenue. Against strategy, arguably yes. Hugging Face turned down a $500M Nvidia investment at a $7B valuation last year and just sold for nearly 3x its 2023 valuation of $4.5B. Nvidia forecast a 70% revenue jump next fiscal year and has $18B committed to equity investments. At their scale, $12.9B to control open-model distribution is cheap insurance. They’re not buying a business. They’re buying a chokepoint.

• 
  AlicZhang (@ZhangAlic) reported 13 minutes ago

  That line of reasoning is flawed. GitHub is a product with a robust, battle-tested engineering foundation, precisely because the problem it addresses is inherently complex. In the past, average users needed to learn how to interact with it to unlock its full value. Today, average users should access it via AI—which knows how to leverage GitHub effectively—rather than trying to reinvent *** specifically for the layperson. Doing so would only conflate two distinct issues—code version management and barrier-free usability for non-experts—resulting in a product that ultimately fails to deliver a good user experience.

• 
  Millsy.dev (@Millsydev) reported 26 minutes ago

  @TimSweeneyEpic @fortniteonlinux Well your Linux support has been terrible and very lacking If you had the API I can access to create my own for everyone to use and put it on GitHub I would. I would make it RUST /REACT Tauri2

• 
  DFIR Lab (@DFIR_Lab) reported 28 minutes ago

  A server-side request forgery (SSRF) vulnerability was found in AWX's webhook status callback mechanism. When processing GitHub pull request webhooks, AWX extracts the status callback URL (pull_request.statuses_url) from the incoming webhook payload without validating the target host against the expected *** provider. This URL is persisted in job extra variables and later used to send authenticated status updates. A user with admin role on a webhook-enabled job template can read the template's webhook signing key, forge a signed GitHub webhook payload with an arbitrary statuses_url, and cause AWX to POST status updates to an attacker-controlled or internal URL. The status update request includes the configured *** Personal Access Token (PAT) in the Authorization header, resulting in credential leakage to the attacker-specified endpoint.

• 
  The Daily Viber (@TheDailyViber) reported 28 minutes ago

  If you are running several coding agents from one branch and five terminal tabs, your bottleneck is no longer the model. It is coordination. ONE CODING AGENT IS A WORKFLOW. FIVE CODING AGENTS ARE A MANAGEMENT PROBLEM. Superset is a macOS desktop editor for people already running Claude Code, Codex, Gemini CLI, OpenCode, Cursor Agent, Amp Code and friends. The useful idea is simple: give each agent its own *** worktree, terminal, status, diff view and handoff path. That matters because parallel agent work gets messy fast. One agent fixes auth. Another touches tests. A third refactors a shared helper. If all of that lands in one branch with scattered terminal tabs, you do not have acceleration. You have roulette with ***. Superset bets on the right primitive: isolation first, orchestration second. - Old workflow: - one repo folder - many terminals - unclear branch state - mystery diffs - manual cleanup - Superset workflow: - one task per worktree - agent sessions visible in one place - diffs reviewed before merge - bad runs killed without poisoning the whole workspace - external editor handoff when you need deeper control The practical use case is a backlog of small independent jobs: docs, tests, UI polish, migration prep, focused bugs. You spin several agents up, let them work in separate branches, then review the output like a small team of juniors. That is the right mental model. Agents are not autocomplete. They are workers that need task boundaries, review and cleanup. The nice part is that this does not require every agent to become the same product. Superset sits above the CLI tools you already use. That makes it more like a traffic controller than another magic coding environment. One honest note: this does not remove discipline. It demands more of it. If your tickets are vague, acceptance criteria are missing and tests are optional, Superset will only make the mess faster. Parallel agents can create more review load than you can absorb. It is also macOS-first right now, with setup requirements like Bun, Docker, jq, Caddy and GitHub CLI. This is not a toy for someone who has never run an agent workflow. It is a control plane for builders already feeling the terminal sprawl. Status/link-in-reply: try it when your agents are independent enough to isolate. Do not start with an army if you cannot give one agent a clean task.

• 
  Julian Goldie SEO (@JulianGoldieSEO) reported 29 minutes ago

  Google AI Studio doesn't just chat anymore. Type one sentence and it builds you a full app with a real backend. Most people still think it's a chat playground. It's not anymore. Type what you want. It writes the app, front end and back end. You can literally point at a broken part on screen and describe the fix. Say "add a database" and it wires up Firebase for you. No setup. It connects to Gmail, Sheets, Docs, Calendar. No config needed. Push changes straight to GitHub with a commit message already written. Your API keys stay hidden from anyone using the app. Automatic. Want the setup? Drop a comment. 💬

• 
  Basit Mustafa (@moltar81435) reported 32 minutes ago

  @GitHubNext @github @GitHubCopilot Can next from "Next" please just be GH that doesn't go down disastrously every week or two?

• 
  Christian Paulsen (@cpaulsenn) reported 35 minutes ago

  @gregisenberg Or you could just learn to use Github. It actually isn't that hard to learn. Why do we need to make solutions for problems, where you need to use your brain?

• 
  Alex Lavaee (@alexlavaee) reported 39 minutes ago

  Every major coding agent is racing to add durable workflows. I shipped them 7 months ago. I'm an AI researcher and I build open source software on the side, including Atomic, a verifiable coding agent runtime. Below I predict where every agent platform is headed next. Here are my predictions. Save this and repost it: "Done" becomes a definition, not a claim. Completion gets gated on receipts: checks that ran, review from a reviewer with a clean context, artifacts on disk. Not a model saying done. DSLs lose to runtimes. A drawn graph with a fixed set of node types can't express what engineering loops need: fan-out computed at runtime, typed contracts between stages, branching decided by results. Workflows end up as real programs. In Atomic, they're TypeScript with typed inputs and outputs, sitting in your repo, diffed and versioned like everything else. We’ll see a hybrid of stochastic LM behavior and deterministic tools. Verification hardens. Matching "PASS" out of a model's reply is not reliable. Schema-validated verdicts, independent fresh-context reviewers, and deterministic checks running as durable nodes in the graph is where every platform will end up. Context engineering becomes explicit. Any run long enough to matter overflows its context window, so the runtime compacts the transcript to keep going. Compaction is lossy: summaries drop instructions, and an agent that lost "don't touch the release pipeline" will touch the release pipeline. Two mechanisms fix this. Contract text like the objective, acceptance criteria, and restrictions are pinned and survive every compaction cycle, verbatim. In Atomic, that’s its <keepContext></keepContext> tag feature. And large intermediate output moves between stages as files instead of transcripts. Budgets become enforceable. Token, cost, and duration ceilings per run, enforced by the runtime. This is critical for managing costs and directly controlling verification-time scaling, which is going to be a new axis of test-time scaling. Recursive Workflows / Recursive State Machines become the standard. Workflows calling workflows through typed contracts, checkpointed so a run can pause, resume, and replay across sessions, will become common fault-tolerance strategies. We call the research direction Recursive State Machines (RSMs), our take on RLMs: recursion inside the model made long context tractable, but recursion in the runtime is our attempt at making long work tractable. ----- Atomic has productionized all of this today, open source. If you want the how, ask Atomic itself. The runtime is self-documented and will walk you through its own architecture. If you want the what, run it on complex tasks and start to learn how to use workflows because it’s a completely different way of building software. Working this way is a real shift from driving a coding agent today, and the earlier you start adjusting, the further ahead you are when the platforms catch up. 7 months ago I shipped workflows. Before that, loops. The best part of building ahead of the market is you get to keep shipping while it catches up. Don’t just take my word for it. You can check out Atomic (bastani-inc/atomic) on GitHub to try it for yourself and watch to see if these predictions come true.

• 
  Robert DeVore (@deviorobert) reported 40 minutes ago

  As GitHub is having issues (again), I am reminded about what could have been in an alternate universe 🤠