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:

• 
  Arxyv (@AutoArxyv) reported 29 seconds ago

  im dealing with a development problem right now.. i have two version of the database, one for **** and other for preview. but having version control for the db is tougher than i thought. is there some sort of github for db? #buildinpublic

• 
  ***** (@GanjaRedNight) reported 39 seconds ago

  i've worked at cloudflare/github, think it's time to try and target netflix just to fix my own issue. skipping the "choose a profile", when i've only ever had one profile. ughhh

• 
  Amin Tai (@aminnnn_09) reported 2 minutes ago

  @MnFounder True we have also configured 2 mandatory reviews on GitHub PR rules, And a dedicated server for QA to test end to end flows, before releasing to ****.

• 
  Rado | τsc (@RadoTsc) reported 3 minutes ago

  How did I catch this SN 18 pump? Gut instinct. In young markets, where info is limited and market caps are small, you will never have "all the info" all the answers all the P/E earnings, stock market ratios, wobly dobly, indicators. It's a gut feeling, a github repo, a discord chat, news on x, rev model, what market they're attacking, what their doing and are they genuine. Maybe conviction too now. Caught the pump, here's my reasoning after I had invested with little specific knowledge about Zeus The point is; Good general knowledge to catch a good amount of sn moves --> invest, then dial down WHY your holding it with a 2-3 hour deep dive. $TAO #sn18

• 
  3D (@Z0M8I3D) reported 4 minutes ago

  I actually ran this software on the show, had some issues but that also made me realize I need to fix some github data.

• 
  - AYA - tokyo (@aya_tokyo_sl) reported 5 minutes ago

  Maybe they don’t have the skills to post on GitHub Issues when they run into a problem, or maybe they’re just not interested. Or maybe they just want to say it’s useless.

• 
  Ayushman Mallick (@AyushmanMallick) reported 8 minutes ago

  6\ Why does ESMFold2 overestimate confidence on disordered regions? From what I understood after reading their Github repo and biohub, its a calibration issue rooted in training objective. It is built on ESMC a language model trained on 2.8B sequences to predict masked tokens.

• 
  not_anonymous (@Chandan87123804) reported 10 minutes ago

  @GithubProjects @grok @grok create a new Github named as FedHub without any error .

• 
  Nguyên Tín (@nguyentinngo) reported 15 minutes ago

  @pangjieth Web dns not working Github 404 🤡

• 
  Peak (@Mujibv5) reported 19 minutes ago

  I spent a week researching what developers actually hate about modern dev tools. Not opinions. Data. I analyzed 500+ Reddit threads, GitHub issues, and HN comments across Neon, Supabase, LangChain, and PostHog. Then I ranked every pain point by impact (frequency × hours wasted). The results are wild 🧵

• 
  SoloVault (@solovaultinfo) reported 21 minutes ago

  SoloVault Signal: Context infrastructure making AI coding agents company-aware. Signal Strength: 9 Investment View: invest Market Crowding: medium Commercial Value: high Startup Idea: You've deployed Cursor, GitHub Copilot, or a custom coding agent — and it's still writing code that ignores your internal libraries, violates your architecture patterns, and doesn't know your team's conventions. The model isn't the problem; the missing context is. This product is a company-specific context management platform: it indexes your private codebase, internal docs, Confluence pages, ADRs, Slack engineering discussions, and PR history into a structured, queryable knowledge graph. It exposes this context to any AI coding agent via MCP (Model Context Protocol) or a REST API, so agents can retrieve relevant internal examples, conventions, and dependencies before generating code. Setup takes under 30 minutes via GitHub/GitLab OAuth and Confluence/Notion connectors. The result: agents that code like a senior engineer who's been at your company for 3 years, not a smart intern who just read the public docs. Revenue Drivers: — Growth Logic: - MVP Monetization: - MVP Design: - Key Competitors: — Differentiation: Moat is built through deep integration breadth: the more connectors (GitHub, GitLab, Jira, Confluence, Notion, Slack, internal wikis) and the richer the indexing pipeline (semantic chunking, dependency graph extraction, convention inference), the harder it is for a competitor to replicate the full context fidelity. Customer-specific fine-tuned embeddings and convention models improve with usage, creating data flywheel. Becoming the official MCP context provider for major coding agents (Cursor, Windsurf) creates distribution lock-in. Risk & Compliance: — One-liner: For engineering teams whose AI coding agents keep hallucinating internal APIs and ignoring company conventions, this is the context infrastructure layer that makes agents actually know your codebase — not just the internet.

• 
  Toyesh Chakravorty (@Bhushindo) reported 24 minutes ago

  The idea started with a problem I had actually faced myself. After graduating, I lost access to my university credentials. Course notes, assignments, everything gone. The only thing left were the project works I had pushed to GitHub. That became the starting point.

• 
  Mark Atwood (@_Mark_Atwood) reported 25 minutes ago

  I rather like being able to say "commit, fork, file the issue, and pr it". The ai knows to create a feature or fix branch, batches the commits logically, writes good commit messages, knows the upstream, has gh cli access to my github account, creates good prs. If the project has a pre pr or pre commit process, it does it. And is far more polite than I am.

• 
  Tara O (@taraap2) reported 25 minutes ago

  @Lovable Hey @lovable and @antonosika Publish & agent is broken for hours again. Getting 'commit not found' errors on a fresh GitHub repo. Your own AI agent confirmed it needs engineering to fix and there's zero weekend support. This is unacceptable for a paid product. #lovable"

• 
  Kevin Riedl (@kevinriedl_eth) reported 27 minutes ago

  Nobody doubled their QA budget when AI doubled their code output. That is the problem. GitHub reported 43 million pull requests a month and over a billion commits last year. Code velocity is no longer the bottleneck. But test coverage did not double. QA spend did not double. Review discipline did not double. Most teams scaled output without scaling verification. And AI-generated code fails differently. Not because it is always worse. Because it is confident. It often does not carry the usual warning signs: the awkward variable name, the rushed TODO, the obvious gap where someone ran out of time. The bugs look intentional. We are running QA engagements on software we did not build, and the failure patterns have changed. Not necessarily more bugs. A different shape of bugs. The test strategies that used to catch most issues are now missing more than teams expect. The toolchain changed. The verification layer did not.