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:

• 
  Blaze (@browomo) reported 4 minutes ago

  This 13-year-old Thai student solves Codeforces rating 800 problems in C++ in 45 seconds through an AI agent he built himself on Claude Code and posted to an open GitHub repo. He sits in a regular school room with a MacBook Air on the desk, a silent HHKB Type-S keyboard for $300, and a timer in the frame. In the browser Codeforces is open, in VS Code an empty .cpp file, and in the corner of the screen a Claude Code window. No algorithm tutor, no competitive programming course, and not a single line of code written by hand. His entire loop runs on 3 components: 1/ Claude Code as the brain 2/ An MCP plugin for Chrome that reads the problem statement right from the Codeforces page 3/ A public GitHub repository where his entire agent lives. Over the past month he closed 23 problems in virtual contest mode. Average solve time, 45 seconds from opening to Accepted. Number of lines he typed by hand during that time, 0. The Codeforces rating system brought his profile to 800 in 12 days. The entire setup is assembled from his repository in 1 weekend. And here is the system prompt he gives the Claude agent before every session: "you are a competitive programming agent for Codeforces. your tools: read_problem(url), generate_solution(language, constraints), validate_against_examples(input, expected), submit_to_codeforces(code). any problem fed to you by the MCP plugin from the browser goes through 4 steps: 1) read the statement and identify input/output format 2) determine the algorithmic pattern (sorting, math, greedy, dp, graphs) 3) generate a solution in C++17 using stdc++.h 4) run it against examples before submitting. if even 1 example fails, rewrite the solution and repeat validation. language: only C++17. file: always {problem_id}.cpp" Meaning the agent knows exactly which problem it is solving right now. It knows what input format Codeforces expects in this specific problem. It knows which algorithmic pattern to apply based on the 800 rating and the tags in the statement. It knows it has just 45 seconds before the student hits submit, and that any failure on a test example is its problem, not the kid's. The work cycle fits into 1 turn of the palm. The guy opens problem 1971A on Codeforces, the MCP plugin picks up the statement, Claude in 8 seconds returns ready code with include, a while loop, and cout output. The guy does a copy-paste into the .cpp file, hits submit, and 30 seconds later sees green Accepted in Status. And only if Codeforces returns Wrong answer on test 2 does the agent automatically grab the failing input, break down the edge case, and rewrite the solution before the guy manages to close the tab. Here is what his Status page shows for the last 24 hours: "1971A - My First Sorting Problem | C++17 | Accepted | 45 sec" "1850A - To My Critics | C++17 | Accepted | 38 sec" "1807A - Plus or Minus | C++17 | Accepted | 41 sec" "1791A - Codeforces Checking | C++17 | Accepted | 52 sec" "1676A - Lucky? | C++17 | Accepted | 33 sec" On the desk in front of him is a MacBook Air without a single open notebook, a silent HHKB Type-S on which not 1 key was pressed during the entire session except cmd+v and cmd+enter, a mi timer set to 45 seconds, and souvenir Zelda figurines behind the monitor. Out of everything I have seen this year, this is the cleanest competitive programming pipeline on a single student: 23 problems solved in a month, 45 seconds each, and 0 lines of code written by hand.

• 
  Manthan Gupta (@manthanguptaa) reported 13 minutes ago

  There have been enough GitHub outages and degraded incidents recently that it’s starting to feel less like random blips and more like a pattern. At this point, GitHub isn’t just “a tool”, it’s the core infrastructure for the entire software ecosystem. When it slows down or goes down, it directly blocks development workflows across thousands of teams. CI pipelines stall, PRs freeze, and releases get delayed. It becomes a bottleneck. That’s why moves like Ghostty stepping away from GitHub are interesting. Not necessarily because GitHub is "broken", but because it highlights a deeper concern and over reliance on a single centralized platform. And I don’t think this is just “scale is hard”. GitHub has operated at a massive scale for years. The frequency of issues lately feels like something else. Maybe architectural constraints, maybe internal changes, maybe just the growing complexity of what GitHub now does. Either way, when your entire dev loop depends on one platform, even small reliability issues start compounding into friction. If this continues, I wouldn’t be surprised if more projects start exploring alternatives, not out of preference, but out of necessity.

• 
  CryptoD₿S (@DbsCrypto) reported 15 minutes ago

  The loudest feedback is usually the cheapest. I care less about the clever Reddit comment than the 3rd confused GitHub issue from someone who already installed, integrated, and got stuck. Public feedback matters when it comes with cost. Time spent. Setup done. Real friction hit. Everyone else is mostly writing fanfic for your roadmap.

• 
  KiwiNod (@Kiwi_Nod) reported 21 minutes ago

  @OludayoDamilol @pharos_network Front-end race conditions on mobile? Okay, that's a real attack vector most auditors miss. 📱 But "logic" isn't a GitHub repo. Show me a write-up, a PoC, or even a thread breaking down a specific exploit you found. What dApp did you break? Where's the teardown? 🥝

• 
  Kush (@kushbhuwalka) reported 23 minutes ago

  I don’t really get the GitHub hate. It works well for me and it’s easy to use. It’s basically free. Last week was the first time I ever saw it down, and frankly that didn’t affect me much either.

• 
  David Zhang (▲) (@dazhengzhang) reported 25 minutes ago

  "Don't write another kanban board" they said Guess who made another board for agents? I got tired of maintaining knowledge wikis in github, and tired of copy pasting context between different agents You wouldn't force your remote team to all DM you right? But that's how it felt working with all my agents If this is a problem you've felt, lmk, will share early access soon

• 
  Human Layer (@HumanLayerlabs) reported 25 minutes ago

  The fix is a score, not a checkbox. A wallet with 3 years of history, 200 GitHub commits, and real on-chain activity scores 82. A wallet created last week scores 12. Same rules. Completely different outcomes.

• 
  Mick.net - Saas 💻 & Aviation ✈️ (@mick__net) reported 27 minutes ago

  @SamuelBeek I just tasked codex to spin up a Azure Windows VM and install my app there to test it, then i use the ‘Windows’ Osx app to remote login and test. Using a Github runner to build the Windows x64 version from my Mac

• 
  Prajwal Tomar (@PrajwalTomar_) reported 29 minutes ago

  Most people using Claude are wasting HOURS re-explaining the same thing every session. This CLAUDE .md hit #1 on GitHub with 82K stars and most Claude users still don't know it exists. This permanently fixes the repetition problem.

• 
  Aryan Sawant (@aryan97208) reported 29 minutes ago

  If you’re a Computer Science student, read this: Stop wasting time on 10 courses. Do this instead 👇 1. Learn 1 language properly (Python/Java/C++) 2. Build 3 real projects (not tutorials) 3. Learn *** + GitHub 4. Solve 1–2 DSA problems daily 5. Start applying before you feel “ready” That’s it. No one cares about your certificates. They care about what you can build. Start today.

• 
  OTIS (@mtende_kuyokwa) reported 29 minutes ago

  @chimze_art I don't even want to fix them right now because of exams lol. I just used it as an opportunity to play with GitHub project and create a kanban table. What do you use for login? Username-password or social login

• 
  BadKidsEnjoyer (@BadKidsEnjoyer) reported 30 minutes ago

  SN62 @ridges_ai went parabolic → afterwards we corrected hard But the Team never stopped building While price dipped: -Ridgeline live: autonomous agents that solve GitHub issues end-to-end -Harbor integration + multi-language evals -Dynamic screeners, real-time patches, commit stats -Scoring 73-88% SWE-Bench Verified + 96.3% Polyglot Hard Facts: -Market cap: ~$32M (FDV ~$145M) -Built a Cursor/Claude competitor with just ~$10M in TAO emissions -Cursor sits at $29B valuation Product accelerating ✅ MCap still tiny✅ $TAO #SN62 #RidgesAI

• 
  Ayush Sharma (@theayush) reported 30 minutes ago

  @peer_rich I am optimistic that GitHub will fix it before any solid alternative arrives

• 
  eric (@ericlim) reported 33 minutes ago

  @domirosari0 @duolingo please fix github

• 
  I Pun Daddy (@IPunDaddy) reported 35 minutes ago

  GitHub down again? Who else is suffering? #github