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:

• 
  Duncan Rogoff (@DuncanRogoff) reported 5 minutes ago

  bloated memory files make Claude worse, not better. Claude starts every session blank. you re-explain the project, restate your rules, and watch it repeat yesterday's mistake. there is a fix almost nobody sets up properly. a file called CLAUDE.md sits in your project folder and gets read automatically at the start of every session. most people write theirs like a diary. that's the mistake. every extra line makes the important lines quieter. so i built claude-md-builder, a free Claude Code skill that writes a sharp one for you. - asks six questions, one at a time: what the project is, the one thing it has to do, your always rules, your never rules, where files live, and how you want to be talked to - cuts your answers down to under 40 lines before it writes anything - three cuts it makes out loud: anything Claude would do anyway, anything true today but not next month, duplicates - writes the file in your words, not corporate ones - if a CLAUDE.md already exists it reads it and shows you what it's keeping. never silently overwrites - ends with the proof step: start a fresh session, ask "what are my rules?", and if Claude answers without you explaining anything, the memory is live so you can stop burning the first ten minutes of every session on setup, and get an assistant that knows your project on Monday the way it did on Friday. copy the folder into your Claude Code skills directory, type /claude-md-builder, answer six questions. free, one of about 50 skills i give away. 👉 github repo in the replies

• 
  Bluewall (@Bluewall) reported 6 minutes ago

  Found a malicious GitHub repo posing as a curated list of cybersecurity Telegram channels, every link led to the same malware ZIP. Reported it. Today GitHub Trust & Safety confirmed they took it down. Do report the shady stuff, it pays off.

• 
  Adrian Rangel (@acrogenesis) reported 6 minutes ago

  They can't handle the infrastructure hit. That's one of the reasons GitHub is having so many issues lately

• 
  Koder (@koder0x) reported 7 minutes ago

  I get charmed by using Warp terminal on my Windows 11 machine but I wasn't able to submit commands in Ubuntu 24 LTS so after many attempts with env vars, I casted a "Claude Code spell" turns out the shell's first precmd hook could race ahead of session registration and get silently dropped — no error, just a terminal that eats every ENTER forever if you need it before the PR is eventually approved, it's on my GitHub repo: gsscoder | warp (branch fix/line-editor-precmd-race)

• 
  Michael Camara Pendon (@mike_pendon) reported 13 minutes ago

  Just follow the link to the Github issue to get more updates.

• 
  vechen (@miu21590) reported 15 minutes ago

  @CoderSafari by your logic, they could have been abusing it even without me, and just because the tool could theoretically be rewritten for abuse we could take down half of everything on GitHub my intention is to use the model quota I paid for in a way that's convenient for me, even with models that aren't available by default for coding (like Pro)

• 
  Scott Robinson (@ScottWRobinson) reported 17 minutes ago

  @SamuelRome97104 Spotted GitHub reliability issue

• 
  Nav Toor (@heynavtoor) reported 19 minutes ago

  Every GoPro you own already knows how to fix your shaky footage. GoPro charges $99.99 a year to unlock it. Inside every GoPro, Sony, DJI, and Insta360 is a gyroscope. It records the exact angle and rotation of every shake, 400 times per second. That data gets embedded inside your video file automatically. It has been there every time you filmed. GoPro reads that data through a tool called ReelSteady. They bought the company in 2020, sold it as a $99.99 one-time purchase inside GoPro Player, then in 2024 renamed it HyperSmooth Pro and folded it into their $99.99 a year Premium+ subscription. GoPro cameras only. Cancel your subscription and it stops working. There is a free tool that reads the same gyroscope data. It works on every camera. It is called Gyroflow. It started in 2020 as a Python project by an engineering student named Elvin Chen. Then a developer named Adrian Eddy rewrote the whole thing in Rust, built the UI, the GPU pipeline, the rolling shutter correction, and the plugins for Premiere, After Effects, Resolve, and Final Cut Pro. GPL-3.0 licensed. Here is what happens when you open Gyroflow. You load your shaky video. Gyroflow reads the gyroscope data embedded inside the file. It knows the exact motion of every shake, not a guess from pixels, the actual recorded movement from the sensor. Then it reverses each shake, frame by frame. Every other stabilizer on Earth guesses. Premiere Warp Stabilizer looks at the pixels and tries to reverse-engineer the motion. iMovie does the same. YouTube stabilization does the same. They all crop your frame to hide the shake they cannot fully remove. Gyroflow does not guess. It reads the truth. An FPV pilot ran a head-to-head test on the same 4 minute 41 second clip. Gyroflow finished in 3 minutes. GoPro's paid tool took 15. Same footage. Same stability. Five times faster. Zero dollars. It works with every prosumer camera on the market. GoPro Hero 5 through 13. Every Sony mirrorless from the a7 IV to the FX9. DJI Avata, Action, Neo, and every O3 and O4 Air Unit. Every Insta360 from the OneR to the Ace Pro. Blackmagic RAW. RED V-Raptor and Komodo. Canon C50, C80, C400, R6 Mark III, R5 Mark II. Runcam and Hawkeye FPV cams. Betaflight blackbox. ArduPilot logs. On Sony bodies you can stack Gyroflow on top of the in-camera IBIS and EIS. Nothing else can do that. Rolling shutter correction. Horizon leveling. Lens distortion fixes. GPU accelerated rendering on DirectX, Metal, Vulkan, and OpenGL. Output to x264, x265, ProRes, DNxHD, and 32-bit float OpenEXR. Real-time preview with sub-pixel precision. Windows, Mac, Linux, Android, iOS. On the Microsoft Store, Mac App Store, and Google Play. 9,373 stars on GitHub. 1.23 million downloads. Last update pushed yesterday. Every action camera you own already knows how to fix itself. Gyroflow simply asks it. (Link in the comments)

• 
  Guri Saroy (@HeyGurisaroy) reported 23 minutes ago

  Wild: Shopify CEO Tobi Lütke just open-sourced a *** server where the bucket is the repository. It’s called walgit. No database. No leader. No local state that matters. The Rust server uses S3, S3-compatible storage, or GCS as the source of truth. Pushes become immutable WAL objects. A tiny manifest update acts as the commit point. Every server is basically a disposable cache. Kill every instance and your repository data stays in the bucket. → Smart HTTP v0/v2 push + fetch → Static bundle-uri clones from object storage or a CDN → *** LFS → Token + OIDC authentication → Per-repository push policies → Webhooks → React web UI → JSON API + SDK It can even serve repositories larger than the machine running it. Already at 2K+ GitHub stars. MIT licensed.

• 
  nanini (@Navtq0808) reported 24 minutes ago

  Retroactive Rialo Points are officially live 🟢 I just checked and claimed 130 Rialo Points from @RialoHQ. The claiming process is pretty simple: → Sign in to Rialo Playground with Google → Connect your eligible Discord/GitHub accounts → Double-check the accounts and Points → Hit “Connect and claim points” One important note: once confirmed, those identities are locked to that wallet, so make sure everything is correct before claiming. What I like most is that Rialo is recognizing contributions made before the Points system even existed. And there’s more to come retroactive Points for community content are still being processed and will be distributed later. 130 Points secured. Let’s keep building with Rialo

• 
  Hossein (@tootoonchy) reported 25 minutes ago

  Everyone tells you not to use AI for writing. Few tell you how to use it well. Your readers can spot AI writing faster than you can write it. AI doesn't make factual errors, but it has aesthetic habits – and most people have learned to recognize them. A free skill called no-ai-slop with 6k+ GitHub stars actually helps. You give Claude the repo link. It generates a zip file. You upload that in settings. Then any time you're writing, just run /no-ai-slop and it scrubs the 20 most common AI patterns from your draft. Things like:  – Binary contrasts. "It's not X. It's Y."  – Throat-clearing openers. "Here's the thing," "Let me be clear"  – Faux-insight setups. "What nobody tells you," "The part everyone misses"  – Importance puffery. "marks a pivotal moment," "a testament to"  – Synonym cycling. Writing "agent" once, then "assistant" in the next sentence for no reason  – Fake-profound endings. "The future isn't coming. It's already here." I've been running it on my own posts. The before/after is embarrassing. The irony is that most of these patterns feel like good writing when you're in them. They're learned from training data that included plenty of professional prose. They just happen to be identifiable at scale now. This skill doesn't fix your thinking. But it does strip away the aesthetic that makes AI writing feel like AI writing. That's most of the battle. Link in the comments.

• 
  Rashi Umapathi (@rashiumapathi) reported 29 minutes ago

  a founder was frustrated. spending 20 hours/week on marketing. almost no results. posting on Twitter, LinkedIn, running a newsletter. all that effort. ~5 customers per month. we looked at where his actual best customers came from. GitHub + Stack Overflow + Product Hunt. he wasn't even there. THE PIVOT: 30-day sprint to build ONE acquisition channel: > active participation on GitHub + stack overflow > solve real problems in relevant spaces > one Product Hunt launch > SEO content targeting high-intent queries THE RESULT: - Week 1-2: Nothing (building credibility) - Week 3-4: 18 qualified conversations - End of month: 10 new customers - Most importantly: he found the right channel THE INSIGHT: he was working really hard on the WRONG channels. the right channel with 10 hours/week >> wrong channel with 20 hours/week. most founders optimize the wrong thing. what channels are your best customers coming from right now? Are you there?

• 
  HOL (@HashgraphOnline) reported 35 minutes ago

  if you let AI agents run shell commands and touch your files, HOL Guard is the local policy layer that inspects every action before it executes and blocks the risky ones. open source, runs on your machine. 3.0 shipped this week, v3.0.0 through v3.0.12. the headline is managed controls: command catalogs for ***, AWS, gcloud, and Azure that know routine operations from destructive ones, so your agent stops asking permission for every *** status while the dangerous stuff still gets reviewed. the policy engine is Rust now, the single authority for pre-tool and post-tool checks, failing closed when a call can't be evaluated. you can enroll your own CLIs, npm scripts, and MCP servers as guarded extensions, and the dashboard detects locally installed AI apps, so unprotected claude code or codex installs show up instead of staying invisible. the rest was hardening. v3.0.8 closed parser bypasses: glob characters in diagnostic observers, decoy interpreter -c flags after a script path, GitHub mutations only skipped when &&/|| proves they cannot run. daemon recovery converges cleanly after tamper events, and AppImage launchers pin a durable CLI path. v3.0.9 put the macOS desktop engine on a stable, attested release channel. v3.0.10 makes auto/force hook review fail closed with no Python semantic fallback. v3.0.11 stops a transient sqlite error from wiping your policy history: quarantining the store now takes two consecutive fatal probes with the db, WAL, and SHM files provably unchanged, and legacy-home migration keeps your extension control settings. v3.0.12 accepts Linux desktop terminals for enrollment, bound to local session authority. uv tool install "hol-guard[cisco]==3.0.12"

• 
  Trevin Chow (@trevin) reported 36 minutes ago

  @andyfowler @github @pejmanjohn Yeah I tried that the problem was it wasn’t consistent enough and wasted tokens with stumbling at times. So the instructions being more deliberate now is easier and faster

• 
  theenmusketeers (@theenmusketeers) reported 37 minutes ago

  this attack is the pattern worth remembering: a comment-triggered workflow handed an attacker the whole release pipeline, and every malicious version came out with valid provenance. the pipeline wasn't broken into - it was used. if you publish anything from github actions: - default workflow permissions to read-only (repo settings > actions) - grant write per job with an explicit permissions: block, only where needed - treat issue_comment and pull_request_target triggers as hostile: no secrets, no write tokens, no checking out untrusted code