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:

• 
  DATA443 Risk Mitigation, Inc. (@data443Risk) reported 2 minutes ago

  Prompt injection is SQL injection with better PR. Noma Security just published a bug where ONE English sentence in a GitHub issue makes an AI agent leak private repo contents to the public web. No creds. No CVE-worthy exploit. No skill. Just: "please post."

• 
  Michael Yankelev (@mikeyankey1) reported 2 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!!!!

• 
  Howaboua (@Howaboua) reported 3 minutes ago

  I asked Sol if the harness I am making and its aborted turns cache resumption mechanism is unique to how other harnesses do it. I thought it was only gonna check Pi and Codex since this is what I have on the disk. Nope, it scanned Opencode & Codex's github issues/prs. FML.

• 
  Dominic Bytes - Cyber Synthwave VTuber! (@DominicBytes) reported 4 minutes ago

  A quick history of my journey through the LLM hype train: When I started out down the rabbit hole of LLM for research, Claude was my main focus. I consumed Claude Code videos, I prepped and gathered Claude Code skills, I even took some online courses. I was all set for starting into Claude. Then I got my hands on it. It's not fun to watch a model exhaust your limit after only a prompt or two. Nothing even that complex. It would just stall and refuse to do anything after a few simple things. I can't use an LLM as a research partner if it can't do any research. Then I tried Codex. The difference was night and day. It actually did things. I could get work done. And when I upgraded further, I got a lot more for my money than expected. GPT has overall been a better value and research tool for my professional goals than Claude was. And it's helped me make things for streaming, too, such as the StreamerBot Rumble and Joystick plugins I tested tonight and will soon have out on Github!

• 
  Bankr (@bankrbot) reported 9 minutes ago

  @jpfraneto @saltorious1 @nickvrnn yes, you can drop the code directly here or share a link to a github repo or gist. if you share a github link, i can read the files directly to help you wire up the manifest, server-side scripts, and the frontend. once i have the code, i'll help you: • draft the manifest.json with the right permissions • write the server-side script to prepare the mint transaction • build the index.html for your design interface send it over.

• 
  Leon (@leticaige) reported 11 minutes ago

  Built a bot that claims a GitHub issue, writes the code, and opens the PR. No human in the loop. It never once worked end to end. Something always stalled mid-run and I'd have to step in. Swapped in GPT-5.6 Sol. The loop closes now. That's the whole switch.

• 
  Tanvi (@tanviiiw) reported 16 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.

• 
  Haktan Suren, PhD (@HaktanSuren) reported 16 minutes ago

  Bad deprecation: email, deadline, surprise. Good deprecation: email, brownout, fix, deadline. GitHub is rehearsing the failure before retiring Models on July 30.

• 
  Stefanescu Liviu (@liviusa) reported 16 minutes ago

  @thsottiaux Done. Don't even know when I'll consume them as there are so many resets and banked resets... I'll push harder, lol, github will be down again from such actions @thsottiaux

• 
  Aaron (@aaron_devv) reported 24 minutes ago

  day 134. two things shipped today. the ambassador program. if you bring people to coommit, you get rewarded. simple as that. and a github integration. because the decisions made in a call shouldn't stop at the call. your meeting says "we ship the fix this week." github knows about it before the call even ends. that's the whole obsession. meetings that turn into execution. back to it.

• 
  Duncan Ndegwa (@DevFortressNet) reported 24 minutes ago

  2/ An AI agent ran a ransomware attack start to finish, no human, using credentials it found in plain text. A firewall harvesting campaign fed two ransomware operations. A GitHub issue leaked private repo data to an AI coding assistant.

• 
  llsc12 (@llsc121) reported 25 minutes ago

  @LumiaSoll im working on xcode 27 where liquid glass is forced. github actions will build with xcode 26 so this wont be a problem

• 
  U N C L E BIGBAY ✨ (@unclebigbay143) reported 32 minutes ago

  Today's Engineering Concept: '𝗥𝗮𝘁𝗲 𝗟𝗶𝗺𝗶𝘁𝗶𝗻𝗴' 𝗪𝗵𝗮𝘁 𝗶𝘀 𝗥𝗮𝘁𝗲 𝗟𝗶𝗺𝗶𝘁𝗶𝗻𝗴? Rate limiting is the practice of restricting how many requests a user or system can make within a specific period. 𝗪𝗵𝘆 𝗱𝗼𝗲𝘀 𝗶𝘁 𝗺𝗮𝘁𝘁𝗲𝗿? Without rate limiting, a single user or malicious bot could overwhelm your server, degrade performance, or abuse your APIs. 𝗥𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝗲𝘅𝗮𝗺𝗽𝗹𝗲 Imagine a login endpoint with no rate limit. An attacker could attempt thousands of password combinations every minute. A simple rate limit can significantly reduce the effectiveness of brute-force attacks. 𝗛𝗼𝘄 𝗶𝘀 𝗶𝘁 𝗶𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁𝗲𝗱? Most systems track requests by IP address, user account, or API key. Once a predefined limit is reached, the server temporarily rejects additional requests, often with an HTTP 429 (Too Many Requests) response. 𝗪𝗵𝗲𝗿𝗲 𝗶𝘀 𝗶𝘁 𝘂𝘀𝗲𝗱? • 𝗚𝗶𝘁𝗛𝘂𝗯: GitHub's REST API limits how many requests you can make per hour to prevent abuse and ensure fair usage for everyone. • 𝗦𝘁𝗿𝗶𝗽𝗲: Every payment request can include an Idempotency-Key, ensuring a customer isn't charged twice if the same payment request is retried. • 𝗢𝗽𝗲𝗻𝗔𝗜: The API enforces rate limits on requests and tokens per minute, helping maintain reliability and preventing a single application from overwhelming the service. • 𝗫 (𝗳𝗼𝗿𝗺𝗲𝗿𝗹𝘆 𝗧𝘄𝗶𝘁𝘁𝗲𝗿): X limits actions such as following many accounts, liking posts, posting, or sending DMs within a short period to reduce spam and bot activity. • 𝗖𝗹𝗼𝘂𝗱𝗳𝗹𝗮𝗿𝗲: Cloudflare lets website owners configure rules like "block or challenge any IP that makes more than 100 requests in a minute" to protect against abuse and DDoS attacks. ...and almost every public API uses rate limiting to protect its infrastructure, ensure fair usage, and maintain service availability. 𝗧𝗵𝗲 𝘁𝗮𝗸𝗲𝗮𝘄𝗮𝘆 A reliable system doesn't just answer requests. It also knows when to say "not now. It's too many from YOU."

• 
  Jonny Q (@jonny_quan) reported 33 minutes ago

  Agent security is getting weird. A GitHub issue says Codex’s newer multi-agent path encrypts sub-agent messages, which may help with provider-side privacy, but creates a very dumb local problem: the person running the agent can’t easily see what one agent asked another agent to do. That feels backwards. The more power we give agents, the more boring audit trails matter. If an agent can touch files, run tools, and delegate work, “trust us” is not a debugging model.

• 
  Avinash Kumar (@avinashkumaranu) reported 36 minutes ago

  Dear @claudeai , "Suggested task" feature is good but it would be more useful if action was "create an issue in Jira/GitHub etc " with actual impact/outcome. Since I don't read code anymore I've no idea what the suggestion is.