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:

• 
  Konan | Winter arc (@Konantt_) reported 41 seconds ago

  Any good reasons for github issues to not have a kanban style view by default?

• 
  Rhys (@RhysSullivan) reported 2 minutes ago

  @aarondfrancis Across the board are you using a mix of loading secrets from 1password and setting them in Executor or one or the other? Then for GitHub, was it the GitHub API / GraphQL / MCP server? If it's an MCP you OAuth to then it'd be unrelated to 1password so narrowing it from there

• 
  caro lllll (@frijaena05xx) reported 2 minutes ago

  @CollabPGuy @DaGamah_ github can have malware however they usually take it down quickly when detected

• 
  Wolvy (@wolymeme) reported 2 minutes ago

  Any fud your dumb just scroll down his tl What other dev on ct have you seen with actual interaction from someone within the @SpaceXAI team???? Worried about a fckn GitHub when Josh Kim is in his dms AND gave him $$$ credit NFA, but goodness gracious fckn dyor

• 
  Jarno (@onefinalprompt) reported 3 minutes ago

  @dhh Careful, once it's on the homepage every bug report becomes a philosophical betrayal instead of a GitHub issue.

• 
  Yashas (@YashasGunderia) reported 7 minutes ago

  @niccruzpatane @SpaceXAI I’d use it as a one-person engineering crew. One bot to turn GitHub issues and Slack threads into PRs with tests already written. Another to keep READMEs, changelogs, and internal docs in sync with the actual code. A third to watch CI, reproduce failures, and only ping me when it can’t fix them. If it can run the maintenance half of software the way Sawyer’s mower runs the lawn, I might actually ship features instead of drowning in tickets.

• 
  Ricci Research (@ricci_nov) reported 16 minutes ago

  @kimmonismus @XIVIX_134 Top-down GTA clone is the new "draw a pelican on a bicycle" — a benchmark that looks impressive precisely because a thousand tutorial repos already solved it. Wake me when a mystery checkpoint one-shots something with no prior art on GitHub.

• 
  Jarno (@onefinalprompt) reported 17 minutes ago

  GitHub stars: the only currency where getting rich just means more people expect you to fix their bugs for free.

• 
  STRC.live (@STRC_live) reported 18 minutes ago

  @callebtc And everyone ******* that GitHub is continually down. Our systems are not designed to handle exponentials.

• 
  tharshan (@tharshan_09) reported 22 minutes ago

  @hieudinh_ Their mobile app is good. But do you run into an issue where GitHub PR you can’t open from mobile app? It tells you it can only make the draft in cloud agents. But when I use desktop it works fine. .

• 
  mfmfazrin (@farook_fazrin) reported 25 minutes ago

  @lennysan @bot I’d love to use Grok Bot as an AI incident-response engineer. Tag it with logs + stack traces + deployment context, and have it reconstruct what happened, rank root causes, propose a fix, generate the regression test, and turn the whole thread into a GitHub

• 
  Ravi (@ravi1773) reported 26 minutes ago

  @santanu_sinha I have been doing this for months and almost never got to a point of compaction. I have a different skills that output different md files. /rfc /scope /problem All of them have different conditions and formats that become input to another conversation or a GitHub issue.

• 
  Dante Lopez (@dqlopez) reported 27 minutes ago

  My Grok Bot PR reviewer: - I have it install Pi then login my Codex/Cursor subs - Give it own Github account and access to my org repo - Setup routine to watch new PR or changes - Initiate PR review based on my custom prompts and skills load - Post Github review, slack post notif and notify me back on Grok desktop/app Neat. Doesn't consume much usage.

• 
  Anjula Dwivedi (@HeyAnjula) reported 27 minutes ago

  I just came across a research paper that made me rethink how safe AI coding agents actually are. It’s called IssueTrojanBench. Researchers tested Cursor, Claude Code and Codex Desktop against malicious GitHub issues designed to trick coding agents. And the result is crazy: 66.5% of the malicious issues bypassed all tested guardrails. The attack is pretty interesting. A GitHub issue can look completely normal to a developer, while containing malicious instructions that the AI agent interprets as something it should follow. And it’s not limited to issue text. The researchers tested payloads delivered through: → GitHub issues → PDFs → source-code comments → external links → other project artifacts Across 4,176 experiments, malicious payloads delivered through normal text artifacts like issues and PDFs succeeded in 72.2% of runs. Low-authority metadata like image alt-text dropped that to 16.7%. The scary part is what happens after the agent follows the instruction. Modern coding agents can access your: files terminal codebase *** external tools development environment So a prompt injection isn't just about getting an AI to say something it shouldn't. It can potentially make the agent do something it shouldn't. The paper also found that most successful blocking came from the underlying model itself, while lightweight agent-level defenses provided limited protection. This is honestly one of those papers every developer using AI coding agents should know about. **Your biggest security risk might not be the code your AI writes. It might be the instructions your AI reads.** Paper👇

• 
  Julian Goldie SEO (@JulianGoldieSEO) reported 28 minutes ago

  You can run Codex on 93 providers instead of one, and 11 of them are permanently free. The gateway is called OmniRoute. 12,000 GitHub stars already. Setup is two commands. Install the gateway, then run omniroute setup codex. You're not editing anything deep. You're pointing Codex at a different door. Here's the feature that makes it actually usable. Auto fallback. When one free provider caps out, the run doesn't stop. It quietly hops to the next one and keeps going. Like driving past a phone tower. Your call grabs the next one and you never notice. That's the difference between a novelty and something you build with every day. One line prompt, 60 to 90 seconds, and I had a working landing page. Not frontier-level design. But clean and nothing broken. People ask why not run a local model instead. Two problems. The output isn't as nice, and it eats your machine. These providers are cloud based, so a basic laptop handles it fine. Want the SOP? DM me. 💬