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:

• 
  Peter Skøtt Pedersen (@PeterSkott) reported 1 minute ago

  @_Evan_Boyle @_Evan_Boyle can we have the remote github mcp server work for the github copilot app then?

• 
  Arti | AI Builder (@Artur_roses) reported 2 minutes ago

  Claude Code just closed a GitHub issue, wrote the tests, passed CI, and opened a PR. No human touched the keyboard. This isn't AI autocomplete. The dev loop just got rewritten.

• 
  Jason Bloomer (@JasonABloomer) reported 2 minutes ago

  @yagiznizipli Pffff, what a scam Let me fix your advert; "show us your github so we can scrape all your repos and train our AI on your code, only for any decent ideas you've had to be taken from you and made ours, then handed off to our legal team to crush you." Sorry, I value my work.

• 
  revayz (@0xrevayz) reported 4 minutes ago

  Andrej Karpathy: "90% of Claude's mistakes come from missing context, not a weak model" Without CLAUDE.md the mistake rate is 41%. With proper rules it drops to 3% You don't need a better AI. You need better loops Most people still prompt one task at a time and fix the answers themselves. That means the human is still the loop Boris Cherny from Anthropic said it best: "I don't prompt Claude anymore. My job is to write loops" The shift is simple. Stop giving instructions. Start designing systems that run themselves: Discover -> Plan -> Execute -> Verify -> Iterate until it passes The 6 things that make loops actually work: -Automations that trigger without you -Worktrees so agents don't overwrite each other -Skills that load context instantly -Connectors to real tools like GitHub and Slack -Subagents where the checker is never the maker -Memory so the loop never starts from zero Prompt engineers ask AI for outputs Loop engineers design systems that produce verified outcomes A reliable loop beats a perfect prompt every time Stop being a prompter. Start being the loop engineer

• 
  Jack Wotherspoon (@JackWoth98) reported 9 minutes ago

  @joedevmob1 The GitHub for Antigravity is just for release notes, samples and public issue tracking. It isn't the actual code unfortunately.

• 
  Blum (@Blum_OG) reported 14 minutes ago

  Andrej Karpathy on MCP: "it's a protocol of speaking directly to agents as this new consumer and manipulator of digital information." that is the cleanest way to think about MCP your coding agent is becoming a second worker inside the product it needs the same context you use: repo, docs, browser, database, errors, designs, tickets, payments if you keep pasting those things into chat by hand you are doing integration work manually the best MCP stack for vibe coding: 1. Context7 give the agent current docs this saves you from stale Next.js patterns, old Supabase calls, wrong Stripe webhook shapes, and Vercel config from 2 versions ago 2. GitHub MCP give it the repo, issues, PRs, branches, workflow runs, and review context half of real work lives outside the file you currently have open 3. Playwright MCP give it a browser the agent should click the thing it built, fill the form, check the mobile view, and catch the button that compiles but does nothing 4. Firecrawl MCP give it clean web research use this before building around a third-party API, writing a comparison page, reading changelogs, or checking pricing claims 5. Supabase or Neon MCP give it the database context that matches your stack start read-only. add writes only when you trust the permissions 6. Sentry MCP give it production evidence real stack traces beat "it crashes sometimes" every single time 7. Figma MCP give it design context when the interface matters spacing, layout, copy, components, and screen structure should come from the file, not from a screenshot and hope 8. Linear MCP give it the task queue bugs, feature work, release notes, follow-ups, and PR links belong somewhere more durable than yesterday's chat 9. Stripe MCP give it official payment context checkout, subscriptions, webhooks, billing, and test mode deserve docs close by and human review close behind 10. Filesystem, ***, Memory, Sequential Thinking give it the base layer files, diffs, history, decisions, and longer plans make the agent act like it is working inside a real project recommended install order: 1. Context7, GitHub, Playwright 2. Supabase or Neon, Sentry, Firecrawl 3. Figma, Linear, Stripe when the product needs them 4. Filesystem, ***, Memory, Sequential Thinking as the base

• 
  Akshay Shinde (@ConsciousRide) reported 16 minutes ago

  @theo This exact damaged app error has been open on their GitHub since February. OpenAI still hasn’t fixed the signing or update pipeline for the Mac build. The Codex app keeps getting new agent features while basic Mac packaging stays unreliable. Priorities are obvious.

• 
  Proof of Voice (PoV) (@Proof_Of_Voice) reported 22 minutes ago

  $XDB @XDBchain is a @StellarOrg-fork L1 for branded coins and Web3 payments. PoV by @0xNeodallas:“GitHub has been frozen since 2021.” ✅ Explorer, Laboratory, Atlas dev tools ✅ Gate, Bitget, KuCoin, MEXC listings 🔍 Down 99.99% from ATH 🔍 No audit or bug bounty

• 
  Q Hoang (@0xqwee) reported 22 minutes ago

  I don't think OpenAI's GPT-5.6 surpasses Claude Fable. If it did, it would have resolved all the issues reported in the Codex GitHub repository by now. Atm, only about 10 issues are being resolved per day.

• 
  Deepesh Kalura (@KaluraDeepesh) reported 23 minutes ago

  Filed as GitHub issues: #336: Phone operators need stable unique IDs (not just phone number) #337: Auto-heal sticky assignments when a node dies Future imp task

• 
  Pedro Pellerini (@pepeller) reported 28 minutes ago

  If Mythos/Fable is so great why are there still 8386 open Github issues in Claude Code repository.

• 
  Tymofii Antonenko (@tymofii) reported 33 minutes ago

  @prinseccoo Are you using Claude Code or an MCP server? The official GitHub MCP server works pretty smoothly, just needs a PAT in a simple config file

• 
  Financial Programmer (@RBiancoUS) reported 35 minutes ago

  A dose of reality for end of week. My biggest question is I can't find any reason for the $Gold panic- did they find gold is causing cancer or radioactive? Selling looks like sheer panic. Would you believe someone asks in DM, so how did *you* get so many followers. Then he lets me brew on it for a day and comes back, I was joking do you have a github, presumably to get some code. No wonder I worked alone. I'm challenged socially guess not alone. After a night of 3 scammers one from Nigeria, one Africa. I need to lock dm down or find a way to restrict

• 
  Trace Cohen (@Trace_Cohen) reported 40 minutes ago

  Shipping fast means stuff breaks silently - broken share images, dead links, leaking {{template}} vars, stale content. You find out when someone shares a broken link, not from a test. So I built a 3-part "site health" system that catches it first. The auditor (~200 lines of stdlib Python) fetches my sitemap and, for every page, checks: og:image actually resolves to a real image (entity-decode the URL first — & bit me), <title> exists and isn't a ${template} leak, no {{merge_tags}} or tracking cruft in the visible text, page returns 200 (catches dead routes in the sitemap), and warns on thin content. Outputs a JSON report, exits non-zero on any FAIL. The dashboard — a noindexed /health page that reads that JSON and renders a green/amber/red status, KPIs (audited / clean / warnings / failures), a per-section rollup, and the exact issue on each URL. One glance = "is everything green?" The loop — a GitHub Action runs the auditor 2×/day + on-demand, commits the fresh report (so the dashboard stays live), and fails the run on any FAIL → I get emailed. Find → fix → re-run → confirm green. It even taught me to whitelist false positives ({{firstName}} is legit on a cold-email page). Want your own? Paste this into Claude Code / Cursor — it learns your site first, then builds it for you: Build a site-health system tailored to MY site. Don't assume my structure — learn it first, then fill in the specifics yourself. PHASE 0 — LEARN MY SITE (before writing code): detect my framework/host/layout; find my sitemap; sample ~20-30 live pages across the sections you discover from my URL structure; figure out how my pages set <title>/og:image/meta (static?dynamic OG route? CMS?); identify where my content comes from (hand-written, generated, imported/scraped) — that's where cruft hides. Do a FIRST diagnostic pass and SHOW me what's actually broken vs intentional (broken OG images, dead sitemap routes, leaking {{vars}}/${template}, tracking params, thin pages). Ask me to confirm which "issues" are expected so we whitelist them. PHASE 1 — BUILD IT, customized to what you found: 1) scripts/site-audit.py (stdlib only) — hardcode MY real sitemap URL, MY section names (full-audit the important ones, sample the rest), and MY intentional-pattern whitelist from Phase 0. Check each page for the failure modes you actually observed (OG image resolves to a real image, entity-decode first; title present, no template leak; no leaking merge tags/ad params in visible text; HTTP 200; thin-content warn). Thread-pooled, retry transient errors once, --json report, exit 1 on FAIL. 2) a noindex /health dashboard reading that JSON (status banner, KPIs, per-section rollup, issue list) — match my design system. 3) CI (GitHub Action) — run 2x/day + on-demand, commit the fresh report so the dashboard stays live, fail the run on any FAIL. Then run it once and walk me through the first real report. Build the thing that watches the things.

• 
  Atlantean Gnosis ☀️ (@AtlanteanGnosis) reported 41 minutes ago

  @DionysianAgent When I made an account it said I made it back in 2024, though I don't think I did, is this a glitch or a GitHub thing?