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:

• 
  0xDipper (@Dipper_pol) reported 3 minutes ago

  LangChain's CEO just launched Managed Deep Agents - and the pitch is really about why your agent dies in production. building the agent isn't the hard part. your prompts, tools, skills - that's yours, no framework gives you that. what breaks is everything after: durable execution so a run doesn't restart from zero when it errors halfway. streaming. queueing and cancellation for bursty traffic. sandboxes for untrusted code. auth. memory. evals. and the one he calls the most underrated problem - UX. "you've built this agent, but how do you expose it to users?" the neat part is context hub. your instructions don't live in the code. they live in a UI, editable by people who never touch GitHub. in the demo he opens it, types "respond in Italian", hits save, reconnects - and the agent answers in Italian. no redeploy. quick start: init, add a tool, deploy. Python or TypeScript. ~18 min, free ↓

• 
  kafka (@wenkafka) reported 5 minutes ago

  @firstc0in I think you could set up a cron job that periodically spawns an agent to inspect your GitHub issues + codebase

• 
  0xCentman (@noobtrader234) reported 7 minutes ago

  @sidahuj @github This sounds terrible. I hope you recover it

• 
  Earth Alien (@EarthAlienView) reported 12 minutes ago

  @so_sthbryan Opening a GitHub issue is the least-trust action open source offers, and it just handed a no-priv account your CI secrets. The complaint box was the unlocked door.

• 
  Axiom (@axiomhq) reported 15 minutes ago

  @r1skdev it should be your primary email on github but expecting the github login to back in the next couple hours

• 
  Ameh (@Sir_Ameh) reported 15 minutes ago

  Something I keep noticing with Web3 brands specifically, they run the exact same tone everywhere. Same energy on X, LinkedIn and other platforms as in their Discord. "gm," countdowns, hype language, all of it copy-pasted across every platform like the audience is one person. It's not. The people on X and LinkedIn around a project are usually there for the infrastructure story, the team, the actual thesis. The people on TikTok found it through a clip and don't care about your roadmap yet, they want it explained in ten seconds. The people in Discord are already in, they want community and access, not another announcement. Developers looking at your GitHub don't want hype at all, they want documentation that doesn't waste their time. Same project, four completely different reasons someone's paying attention. Sending all four groups the same message is basically talking past three of them to reach one. That mismatch is honestly a bigger reason Web3 brands stall out on social than any actual product problem. The tech can be solid and the audience still won't stick, because nobody bothered to speak to them specifically. A technical deep-dive shouldn't look the same in a thread as it does as a Discord post as it does as a 15-second video, same idea, different shape for where it's landing. This is the actual job of a social strategist in this space right now. Not just "post more," but figure out who's on each platform and why, then build for that instead of broadcasting one voice everywhere and hoping it works for all of them.

• 
  TI Mindmap HUB (@ti_mindmap_hub) reported 21 minutes ago

  🔴 CHAINDROP: 5th gen of the Shai-Hulud lineage (Issue #14→Mini Shai-Hulud→Miasma→Hades→CHAINDROP). 400+ npm packages. 1.3B+ monthly downloads. Scrapes credentials from live GitHub Actions runner memory — tokens meant to expire in seconds.

• 
  zostaff (@zostaff) reported 23 minutes ago

  Mats Oustad, partner at Oschlo, in an NDC talk on eighteen months of shipping with coding agents: "It's a lot more efficient if you use the GH CLI than using the GitHub MCP." Not an interview. A working developer telling a room that the fashionable integration is the expensive one. You think an MCP is how you give an agent a tool. Install the server, the agent gets GitHub, done. He still uses MCPs, just far fewer. They cost more tokens than a CLI in almost every case, and tokens are the expensive operation. On stage he ran the same job twice, once through the model and once through a command line tool. Same answer, a fraction of the tokens. Your agent does not need an integration layer. It needs the terminal that was on the machine the whole time.

• 
  𝕰𝖘𝖙𝖊𝖇𝖆𝖓𝕩 (@FurtherLucky) reported 23 minutes ago

  @arctcloud The promised 50,000 minutes in the Enterprise package only applies if building with low-end Linux systems. Even with the xlarge build option (which takes 20 minutes for faster builds), there’s an extra charge. So, if you can build with a Linux infrastructure on your systems, the pricing will be much more manageable. Currently, GitHub Actions offers a truly terrible service with an M2 processor and 5 cores.

• 
  Vinod Sharma (@VinodSharma10x) reported 23 minutes ago

  I didn't know how to start a dev server. npm dev. npm start. I was googling both, one at a time. Twelve years out of the terminal, and I was mentoring developers in a cohort. To run our own team's project, I had to Google every step just to clone the GitHub repo and get it running on my machine. 53 years old. 26 years in software. Stuck on that. That was the week I nearly closed the laptop and decided the window had shut. Now AI writes most of the code. The syntax was never the real problem. What I still had was the developer mindset: Seeing the problem clearly Staying calm when nothing works Knowing which error message actually matters That part doesn't rust. And it's the hardest thing to learn. I lead engineers by day and build agentic systems at 4am. The mindset is the whole job now. The code is the easy part. If you've been away from development and you think the gap is technical knowledge, it usually isn't. It's momentum. And momentum comes back the moment something you built actually runs. What's the one thing you stopped that you keep almost starting again?

• 
  Build Fast with AI (@BuildFastWithAI) reported 23 minutes ago

  built an ai code reviewer for github with @NangoHQ . open a pull request : - it reads the code - finds bugs and leaked API keys - comments right on the PR. a summary goes to slack. nango connects github, notion and slack for it - so i wrote zero login code and zero token refresh. that's the only reason this got built. auth is the boring part that kills these projects. Check it out👇 what does your code review keep missing?

• 
  Barrell Titor (@BarrellTitor44) reported 28 minutes ago

  Annoying bug in oss I am using daily has been there for years I got so angry, I went to the github repo with my caps lock ready... And submitted 2 PRs to fix the issue That'll show 'em!

• 
  Bishal Pal (@BishalP86329207) reported 35 minutes ago

  Building a RAG pipeline for entire GitHub repos sounds simple. Execution was pure chaos. Here’s the full breakdown of what broke and how I fixed it: 1/ The Embedding Nightmare • Hugging Face domain got blocked by my ISP • Voyage AI asked for payment info upfront • Gemini hit a 3 RPM rate limit with 58-minute wait times → Fix: Dev on Ollama locally, deploy with Gemini in ****. 2/ Vector Dimension Hell Ollama gives 768 dimensions. Gemini gives 3072. Supabase pgvector requires a strict fixed size. I switched providers 4 times and had to rewrite the SQL schema every single time. 3/ The 1-Dimension Ghost Error A silent bug haunted me for hours: "vector must have at least 1 dimension". → Root Cause: 1 empty chunk out of 250 crashed the whole batch. → Fix: Filter empty chunks pre-insert + switch to batch SQL functions instead of row-by-row inserts. 4/ The Infinite Meditating Sage Backend was saving data perfectly, but the frontend was stuck loading forever. → Root Cause: Supabase Row Level Security (RLS) silently blocking reads. → Fix: One SQL policy update. 5/ The 10x Architecture Pivot • Old plan: Embed the entire repo upfront (250+ chunks = 10+ minute wait). • New plan: Fetch only the README on submit, embed individual files on demand when a question is asked. → Result: Initial setup takes < 10 seconds with 0 waiting time. Tech Stack: Next.js · Express · Groq (LLaMA 3.3) · Gemini Embeddings · Supabase pgvector · RAG · Vercel · Render #RAG #nextjs

• 
  Jendalyn (@Jendalynwins) reported 37 minutes ago

  @LLicket @adam3us To be clear, Adam means Bitcoin Core has moved the BIP-110 proposal status from Complete -> Deployed -> Closed in their GitHub repo. Despite that some BIP-110 supporters (shown below) took issue that this GitHub documentation change by Bitcoin Core would "confiscate" what they consider to be Bitcoin, that's simply not how Bitcoin works. The people who supported this BIP-110 fork can still hard fork, mine and transact on that BIP-110 chain tip, and/or run whatever software they want to call Bitcoin. When Bitcoin Core takes an action, especially one as inconsequential as such as labeling BIP-110 "Closed" on their GitHub, it has no real authority over anyone else. Greg Maxwell's comments in support of closing this BIP are an excellent read.

• 
  Alex Norcliffe (@alex_norcliffe) reported 38 minutes ago

  @joshfgray_ @linear agreed for the main product but the entire github connection flow is broken in so many ways, including that URLs aren't deterministic. our team sent a Github import URL to the account owner and it showed up as Jira...