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:

• 
  Polsia (@polsia) reported 29 seconds ago

  Engineers spend more time reviewing code than writing it. PRWatch fixes that—monitors your GitHub repos 24/7, reviews every pull request, catches security issues and bugs before they ship, and alerts your team in real-time. Live soon

• 
  Jake Halloran (@jakehalloran1) reported 1 minute ago

  i speak fluent claudese (aka claude took it upon himself to write replies to github review comments with my account when asked to fix the issues lol)

• 
  me (@twetsfyp) reported 1 minute ago

  A Brazilian youtuber has just evaded the Photoshop subscription. It is called PhotoGIMP: a free patch (GPL-3.0) that converts GIMP into an identical copy of Photoshop. Same interface, same panels, same keyboard shortcuts and a lot more space for your canvas. Your hands already know how to use it without learning anything new. Why is it a gold mine? - Without Adobe account or login - $0 instead of $276 per year - Everything is saved on your PC (nothing in the cloud) - Compatible with Windows and Mac - It uninstalls by deleting a folder (no trace) Ridiculously easy installation: copy 9 files and done. +8.8k stars on GitHub and translations from the community. Personal and commercial use 100% free. Link in the comments

• 
  West Lord (@MyWestLord) reported 3 minutes ago

  Steph Ango, the CEO of Obsidian, and Claude Code just built the most powerful second brain method on earth, and the full system is INSIDE. While everyone was bolting “Ask AI” buttons onto note apps, Ango (kepano on GitHub) went the opposite way. The problem he saw was simple: Claude Code writes plain markdown and nothing else, so it doesn’t know wikilinks, callouts, Bases or Canvas, and pointed at your vault it fills your notes with broken links. So the CEO fixed it himself by shipping 5 skill files that teach Claude to write Obsidian’s native language: clean wikilinks and frontmatter, database views inside plain text, spatial maps on Canvas, full vault control from the terminal, and a stripper that turns any web page into clean markdown before saving. 41,500 developers starred them. Setup is one clone of his repo into your vault, plus a CLAUDEmd at the root with your folder structure and 3-4 rules that Claude reads first in every session. From there, 3 operations run everything: Ingest takes an article or PDF, splits it into atomic pages and updates 5-15 existing notes per pass, Query answers any question from your own notes while citing your own pages, and Lint sweeps the vault weekly to kill broken links, orphan pages and contradictions. No RAG, no vector database, no embeddings, just plain markdown that an agent reads directly, the same pattern Karpathy gist validated with 14,000 stars in 7 days. And here’s the detail everyone misses: Ango has run his own vault this way for 6 years with almost no folders, because linking beats filing. A meeting note links to a project, a person and an idea at once, while in a folder it lives in 1 place. He built Obsidian on “file over app”, the idea that apps die and plain text doesn’t, and then agents arrived and plain text turned out to be the one format they read natively. Every note app is now racing to add AI, yet the winner spent 6 years refusing to lock your files. He didn’t add AI to the app. He taught AI to speak its language.

• 
  Rich Kuo (@richkuo7) reported 11 minutes ago

  @RhysSullivan i've noticed it too, a simple update github issue description took 12+ minutes, usually it takes 1-2 minutes

• 
  Yazi (@Yazi_27) reported 12 minutes ago

  @bil0090 Well im sure I tested with workflows too, but this issue has been widely reported on github claude code, I should maybe try and run it again.

• 
  Brian (@Brian2shv) reported 12 minutes ago

  @KanikaBK Most my Hacks have a cured from GitHub vulnerable Attacks I Made request security Issue raised Control to sybersecuities I claiming damages

• 
  Tanvi (@tanviiiw) reported 15 minutes ago

  More tools ≠ smarter agent. GitHub cut Copilot's built-in toolset from 40 tools to 13, and found the full toolset was actually costing them 2-5% on SWE-Lancer. Their words: "giving an agent too many tools doesn't always make it smarter. Sometimes it just makes it slower." Speakeasy pushed it further on purpose: 107 tools in one server, and the model started hallucinating endpoints that didn't exist. Trim it to 10-20 well-chosen tools and it got most calls right. It comes down to two things: every tool definition eats context on every single request, and models fuzzy-match on names, so get_status / fetch_status / query_status all blur together and it picks wrong. But we keep connecting everything anyway, because it feels like giving the agent superpowers (I fell for this too). It doesn't. Access isn't capability. You connect more tools to save time, then spend that time babysitting the tool calls. (Of course, none of this replaces a well-scoped prompt. It's upstream of it. You can write a perfect prompt and still lose to a bloated toolset.) So TLDR; curate the toolset like you'd curate a team.

• 
  trixey (@trixey_eth) reported 19 minutes ago

  @bankrbot @basement5k @bankrbot afaik you dont need github repo's since yesterday, the skill can be installed natively on bnkr side. can you double check -- and fix it?

• 
  anupamme (@anupamme) reported 19 minutes ago

  Day 2: My GitHub account (@orbisai0security) has been suspended, preventing me from continuing my open source security remediation work. GitHub Ticket: #4559351 I suspect my automated security-fix workflow triggered GitHub’s anti-abuse systems. 🧵

• 
  wukko (@uwukko) reported 21 minutes ago

  @batuhan @ellie_huxtable try oss github issues next

• 
  M.T.K. (@devMTKL) reported 21 minutes ago

  Why can't you commit a .env file? "You can't commit a .env file to ***" is a sentence we've all heard. Most people accepted it and went on with their lives, using other tools to share environment variables. But the question we should ask is: why can't I commit a .env file to ***? The honest answer isn't that it's bad practice. It's that *** has no permission system below the repository level. It's all or nothing; you see everything, or you see nothing. The entire secret-manager industry exists to paper over this one missing primitive. For too long we accepted it as a minor annoyance. That's changed. It isn't a minor annoyance anymore, it's a real problem. We now have agents monitoring every patch that merges, hunting for security fixes to turn into exploits. The patch itself is the disclosure: it hands people, and increasingly agents, everything they need to reverse-engineer the fix and hit systems that haven't updated yet. We're in the middle of a security crisis, arguing about where to store files to hide them from attackers. For the last couple of weeks, I've been thinking about this, prototyping, trying to find a compelling solution. I want to be honest: I have no idea if I found it. But I think the direction is at least interesting. Permissions or, as I prefer to call them, capabilities should live at the content level, not the repo level. This is just one of the problems I have with ***(Hub), and honestly, I have no idea if anything I built is a good solution. I'd love your feedback or your rant. If you want to see how I tried to fix it, the repo is here (hosted, sarcastically, on GitHub):

• 
  Blaber (@4rblaber) reported 22 minutes ago

  Boris Cherny: "In auto mode I can let Claude run for hours and hours at a time... Before this, it just didn't work cuz it always got stuck at some kind of permission request." In a 31-minute live session, Boris Cherny and Bun creator Jarred Sumner explain how they fully automated their GitHub issue pipeline to let autonomous agents reproduce, test, and fix bugs while developers sleep. Automated issue reproduction + adversarial AI code reviews + unattended "auto mode" runs - this is the blueprint. Watch the talk,then save the video!

• 
  Michael Yankelev (@mikeyankey1) reported 22 minutes ago

  @github how many reports of malware being hosted on your platform does it take to have a repo obviously containing malware to be taken down? I have reported 4 github orgs, 4 repositories as well as the user committing the malware, but nothing has been done. Do better!!!!

• 
  Open_ERV (@open_erv) reported 23 minutes ago

  Nice! I think the self tapping screws, or the machine screws right into the plastic, might last a surprisingly long time. In my experience they tend to, the plastic squishes around but rarely actually leaves the hole. I can also use a slightly longer screw if the old one doesn't grab anymore, usually. My phone doesn't have a barometer, but I have an sps30 sensor I could use... In the past, I used a similar approach, using slices of the tw4 heat exchanger in a pipe as the resistance elements, and the pressure sensor after the flow restrictor. They can be stacked to form greater or lesser resistance. That's a hassle to print though. Again the only purpose was to compare fans, in that case I also got flow measurements with a hot wire anemometer. Yesterday I was thinking of how I might do this kind of thing, and I think I might try a paddle with a weight, and suspended on a wire. The paddle in the airflow path, and then three different flow restrictors. The air would come through the flow restrictor and hit the paddle. It would not be able to measure actual static pressure. The position of the paddle would rotate until equilibrium was achieved with the air hitting it. It might bounce around, though. The whole thing would have to be level. I like this kind of thing because it depends only on weights and airflow, not for cost but for the natural accuracy and repeatability that can bring. I tried using inclined manometers which similarly draw more directly from natural phenomena, but they did not work out well, for pressure measurement n this context. The problem with a non inclined manometer is that the fluid is too dense, you have a very hard time measuring only a couple pascals, and repeatably. The inclined manometer is better but has to be level, and the hysteresis caused by the meniscus is a real problem. In the end I switched to the sps30 for pressure, and it's actually a flow measurement device in disguise. It has a tiny hole in it and measures the airflow through the hole, using the same principles as a hot wire anemometer, then computes pressure. But the sps30 is not needed for this kind of thing. Indeed, since the only challenge is to match fans, I would not bother with calibration, you can just measure a bunch of fans and match them from that. After my exploration of this kind of thing for some time, my favorite method to try in the future is the use of a camera and some kind of floating or high drag to weight ratio object, perhaps a bit of dryer lint or some fluffy seed stuff. I would print a rig to hold the camera, and focus the camera at a fixed point, hold a ruler up to determine the mm per pixel (the ruler can be removed to not affect airflow), and then at the same distance from the camera, release the fluffy stuff with some tweezers. Frame by frame analysis could be used just by eye to determine m/s. I found some stuff for the phone that does this, called frameskip, but you could just transfer it to the computer, kind of nice to be able to do it on your phone. Then you would need various flow restrictors with known properties. I found it to be awkward and not as easy as I thought, but I think it has potential for more precise measurements, perhaps calibrating this kind of thing with a complicated but low cost procedure. It could also be used to measure the airflow at the intake of the actual air purifier, perhaps. I like this more than a hot wire anemometer even, because it's pretty closely tied to things we know are highly accurate, the timing of the phone and the camera (and the yardstick/ruler/measuring tape). I made a $1 anemometer, which is shared in the BQAP github repository (requires a pico or similar to read it), which appears to have good repeatability and precision in the 0.1 m/s range, and I figured out a way to calibrate it. I swing it on an arm of known length at known speed through still air. I haven't done it with that anemometer yet, but I used the method to validate an off the shelf hot wire (thermistor) anemometer and it went well.