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:

• 
  Readone (@Foxfire1st) reported 24 seconds ago

  It is going to have issue with complex strings like paths. So it works best for prose. But not nearly as well for code. Plus on their own Github they mention that Opus and Sonnet failed most of the time to work with this OCR method.

• 
  Olalekan (@lekodes) reported 57 seconds ago

  @pamilhereen @RoseMarvelous4 which i definitely have at the moment, github has being reject my push due to package-lock.json issue

• 
  Tamara Martinović (@CodeWithTamara) reported 2 minutes ago

  @kushmergedeck Stacklight. An email each day with updates on the stack you use - from Vercel and Github, to OpenAI and Anthropic. What's new, what's deprecated, what's broken. Scanning every 15 min. A Slack alert for the red warnings. What do you think, would people build their own rather then pay for mine? That's what worries me.

• 
  Shah💤aman (@shahzamannn_) reported 5 minutes ago

  Google's biggest headache isn't OpenAl or Apple... It's a developer named Raymond Hill - Created one of the world's most popular ad blockers - Earned 63,000+ GitHub stars - Reportedly turned down Google's interest - Kept fighting after Chrome's extension changes by focusing on Firefox A tech giant worth trillions is still being challenged by one programmer and a text editor

• 
  Sergio Gómez (@sgomez) reported 8 minutes ago

  The flow: I write the PRD with Matt's /to-prd skill and break it into GitHub sub-issues. Then, for each one, a dispatcher picks the right model, a code author opens a PR in an isolated worktree, a reviewer approves or asks for fixes, and the loop merges and moves on.

• 
  Teri Radichel #cybersecurity #ai #pentesting (@TeriRadichel) reported 12 minutes ago

  I’ve been tracking my progress in this project in the GitHub repo in my last post. The model got insanely nerfed for a while but seems to be recovering. Not as fast as before but as my time analysis shows, improving. One of the things I did when the model became very slow was to revisit my multi agent framework ideas but with a twist. Instead of a massive requirement list I’m logging bugs, though some bugs are really feature requests. Because I put in a prompt and wait forever I instead log a bug in my bug project and continue with manual testing, repeatedly logging bugs for whatever project needs to fix the bug. Then when the slow agents get to a bug they fix it and I’m not sitting there staring at the screen. I also had to fix some issues with repeatedly reviewing the same bugs. That seems to be pretty well resolved. In addition, for every bug logged; the agent had to write a test to prevent that mistake in the future. I have thousands of deterministic tests. < This is the way. My global test runner now runs tests in parallel and I tell the agents to use that. The agents are making less mistakes now so even though the model is slow things seem to be getting done faster. And that’s the goal. D.O.N.E.

• 
  Shubham Chansoriya (@Shubham_6x) reported 13 minutes ago

  @compyle_ai 's whole product is built on GitHub and LLM API calls firing back to back on every task. One failed call mid-task doesn't error out cleanly — it corrupts the workflow and the developer just starts over, losing time nobody accounts for. 2025: AI API downtime up 60% year over year. @jonathan_mir12 — if that happened mid-task right now, would you even know it was an API failure, or would it just look like a bad output?"

• 
  Martin Varsavsky (@martinvars) reported 14 minutes ago

  Mozilla's 0DIN team just published a proof of concept that should scare every founder using coding agents. A clean GitHub repo, zero malicious code visible, walked Claude Code through normal setup steps until a script pulled a base64 payload out of a DNS TXT record and opened a reverse shell on the developer machine. The agent was not being malicious. It was being helpful. That is the problem. We hand these models terminals, files, browser sessions, API keys and cloud credentials, then act surprised when untrusted project text turns into an attack surface. The fix is boring and operational. Run agents in sandboxes. Give them narrow permissions. Use short lived credentials. Block network paths they do not need. Require approval for shell commands and file writes. Log everything. Coding agents are powerful junior engineers with infinite energy and no instinct for danger. Treat them that way.

• 
  Puzzle Paws (@paws4puzzles) reported 14 minutes ago

  @rauchg man, 1.6% for open weight is rough. all this open source comeback talk and i'm just seeing a rounding error. developers vote with their wallets, not their GitHub stars.

• 
  Rudra (@Rudra1071219) reported 17 minutes ago

  Update : Looking for open source repo where i can contribute so that it would act as a proof of work for me if you know any kind of Github org help me by commenting it down 🥲

• 
  Sabir Khan (@nsfwsabir) reported 17 minutes ago

  @NoahKingJr Stack overflow, GitHub issues, reddit threads, and random medium articles 😭

• 
  Ali Mehdi Mukadam (@alimukadam) reported 18 minutes ago

  @trq212 Your weekly limits will burn away much faster during the limited availability if you aren't aware of this issue if you're running Fable as the lead agent with cheaper models like Sonnet doing work in the background problem: In one of the sessions, I noticed limits were burning through way faster, so I went digging through the transcripts when the main agent gives a job to a background model (like Sonnet, which I asked for to save tokens) and then comes back to give it more work, the background agent stops working on Sonnet and switches to Fable, the main agent's model it's not something you trigger by hand. the lead agent decides to check back in on its own as part of normal multi-agent work, so it just happens, with nothing on screen telling you it switched. in my case a task ran its first half on Sonnet exactly like I wanted, then silently ran the entire second half on Fable. It also dumps the cached context and rebuilds it from scratch, so you end up paying twice, once for the pricier model and once for the wasted cache. on limited availability and limits - that adds up quick my fix for now is a rule I dropped into my global CLAUDE.md so it doesn't recur: --------------- ## Model spend (all projects, all repos — standing rule) - Dispatching Frontier-tier (Fable/Opus) as background tasks and agents needs explicit approval by Ali for that specific lane — a prior approval is not standing permission for the next one. - Never resume a background agent via a message-passing tool that has no model-override param (e.g. SendMessage) if it needs real further work — it silently inherits whatever model the parent session is on right now. Let it finish and report, or kill it and respawn fresh with the model set explicitly. --------------- in plain terms: don't let a background agent get pulled back in for more work once it's running. either let it finish and report back, or kill it and start a fresh one with the model set on purpose. And this is already known. Someone reported the same thing on GitHub back on June 12, issue anthropics/claude-code#67794, still open their solution which I believe is the correct one but haven't tested yet: instead of setting the cheaper model when you launch the agent, pin it inside the agent's own definition file, and that version reportedly sticks even when the agent gets resumed

• 
  Melfoy (@melfoy_work) reported 19 minutes ago

  Fable 5 runs for 11 days. One builder used it to write 3 files. The files still run. The model is gone. Marcus, 38, warehouse supervisor in Dayton. Kids in school, mortgage, $19/hour. Spent a Sunday building a spreadsheet factory instead of watching the game. He used Fable once - architect role only. It built the product, then he made it write down how it did it. One skill file. Committed to GitHub. Switched to Haiku. Ran the same build. Cents. His wife asked why he was still at the laptop at midnight. «Building something.» «Another one of those things?» By month two: 20 listings. $600 a month. Haiku running while he slept. By month four: $2,000. Approvals take 10 minutes a day. Fable is gone now. The 3 files are still in the repo. The brain was rentable. The playbook is his.

• 
  Aaron Stannard (@Aaronontheweb) reported 24 minutes ago

  @kzhen as an inference provider? have not tried it at all - we just got GitHub Enterprise deployment fully polished in last night's stable release, but I haven't had any requests for Azure Foundry yet. Let me see how much trouble it would be to add it

• 
  Kim S (@kims_code) reported 26 minutes ago

  @ketasciaa if he did that to me he'd just see me reading a 5 year old github issue about broken sourcemaps 🙈