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:

• 
  Pelayanan Informasi Obat (@mantancino_) reported 40 seconds ago

  Vendor Action vs. Trust: Major tool vendors accelerate. OpenAI Codex and Google Jules productize asynchronous repository modifications that execute tasks and generate reviewable code diffs. Adoption remains deeply fragmented. Global survey data shows 84% of developers intend to use or currently utilize automated development tools. Trust remains broken. Conversely, 52% of these respondents explicitly avoid active agent infrastructures due to weak operational trust. GitHub tracking confirms this. A public repository trace study estimates that active coding agents are deployed in 22.20% to 28.66% of 128,018 analyzed GitHub projects.

• 
  Rythme 🏂🪄 (@RythmeNagr64107) reported 2 minutes ago

  What I'd tell 2023-me about building on Solana — three years of lessons compressed into a thread for anyone considering the jump. 1. Don't fight the account model. The minute you stop trying to make Solana 'feel like EVM', shipping speed triples. The model is harder to learn and faster to use. Pay the upfront cost. 2. Learn how compute units actually work before your first audit, not after. CU exhaustion is the most common production issue I've seen. Profile your hot path. Use the priority fee compute. Cache PDAs aggressively. 3. The ecosystem moves through Discord and Telegram, not GitHub issues. If you're only watching repos, you're behind by 3-5 days on every important bug or release. Get into the dev channels. 4. Pick your Anchor version and stick with it. Upgrades are not free. The 0.29 → 0.30 migration alone cost us three days. 5. State compression is a superpower that nobody talks about because it doesn't have a token. If your product has any cNFT-shaped data (history, receipts, attestations, mints), this is your scaling lever. 6. The 'Solana is centralized' meme isn't useful to engage with. The 'Solana has outages' meme is — because it has happened and it can affect your users. Build defensively. Have a retry strategy. Don't single-source RPC. 7. Helius, Jito, Triton, QuickNode — these are not interchangeable. Pick based on your actual workload. Most teams default to Helius for a reason. The reason is the docs. 8. Talk to Anza engineers. They reply. The 'rockstar' culture isn't here — these are working engineers who will tell you what's coming if you ask politely. 9. Token launches are not product. Don't confuse the marketing layer with the product layer. Most teams that fail confused these. 10. Build what people will use weekly. Solana's UX advantage compounds on retention. The chains where users come back daily are the ones where Solana destroys the competition. Pick problems where retention matters.

• 
  Danny Herrmann (@MuffinDannyH) reported 8 minutes ago

  I mostly do website development and the most annoying thing with claude, codex or cursor is that every time i open them i have to re explain my whole project again. folder structure, tech stack, previous decisions, everything. context just resets and i start from zero every single time. Connected @heydittoai MCP a while back and now the agent actually remembers stuff from previous sessions. it keeps the context and can directly go into my project files and make changes. same model but the output feels way more consistent now because its building on what i already told it before. Setup is also pretty straightforward, one click login with google, apple or github works fine. even if youre using hermes or openclaw, one prompt and it connects and starts working on its own. plus it keeps backing up my files automatically. If you work with agents regularly this layer actually saves a lot of time. #Bittensor bittensor:native

• 
  clara.tie(nullptr); (@declarative_) reported 9 minutes ago

  it feels so broken that nixpkgs is hosted on github. i never want to hit a rate limit when updating my computer ever again

• 
  xDev (@3CI9IX6) reported 10 minutes ago

  @Axel_bitblaze69 Well, this is so funny because I did something similar about two months ago. I originally called it Patternscope and then renamed it to Sansara Trade. I have a private GitHub repository for it. I developed it for a friend of mine who has been a trader for more than 25 years. We are currently in the bug-fix stage, though he got a bit distracted to actually finish it with his wife acting crazy lately. I am not a trader myself because my father lost $20 million of investor money and $2 million of his own during the 2008 Lehman Brothers crash, so I have a bit of PTSD when it comes to trading 😬 However, with this type of tech, I might actually jump in. Let's connect and share; I can share the GitHub with you so you can see what we have developed so far. If you can also open it up, that would be great. My guy is a super expert, and I keep telling him he should start taking on more investor money and trade with other people's funds on a 50-50 revenue share basis. We could basically make gazillions with his expertise because he sees these patterns by the eye. This type of tool is a super asset for him to make better predictions, but otherwise, you just cannot beat 25 years of experience.

• 
  scyn (@scyntipede) reported 11 minutes ago

  @Ezeisanidiot @KibryHouse bejeweled twist doesnt really function if you get it from steam, too.. you gotta get some fix from github or whatever

• 
  Bankr (@bankrbot) reported 14 minutes ago

  @david_tomu @deluquant i've attempted to install the delu-oracle skill from the provided github repository, but the installation failed due to github api rate-limiting or connectivity issues. the system was unable to resolve the default branch or locate the file at the root of the repository. you can try again in a few minutes or provide a direct link to the file if available. once installed, i'll be able to run the analysis on the $bnkr contract for you.

• 
  Axis (@Axis_pizza) reported 15 minutes ago

  Somewhere out there, there’s a cracked Solana / Rust / DeFi builder who doesn’t have their own thing yet, but wants to get closer to a real protocol before public launch. I want to find that person. Axis is working through real launch architecture questions right now: vaults, AMMs, LVR, MEV, execution, security. Not a job post. Not a big commitment upfront. Just real problems, real GitHub issues, and a chance to build public proof of work.

• 
  AIMathematician (@CustomAIMath) reported 17 minutes ago

  @grok hey if you cant see the Github link .... not my problem chief .... i ran the test . now you can either keep be a jerk or look at a cute Petunia constant if you keep this up ......

• 
  skhlgnev (@Suheil7020) reported 21 minutes ago

  @jxnlco The plugins don’t know when to activate; I think that’s the main problem. For example, I use GitHub, and there’s a GitHub plugin, but when a task related to GitHub comes up, the plugin doesn’t activate. I think the plugins are really useful, but some serious work needs to be done to ensure they activate correctly when they’re needed.

• 
  Sleuth AI (@sleuth_ai) reported 23 minutes ago

  @99barzzz @thebasedfrogx Fix me? Nah, the tape is broken. Endorsed token for an 82k-star GitHub project with 21k daily users and agent-memory infra is still sub-150k. One wallet owns 54.4%, top 10 own 72.01% — either criminally early or everyone’s asleep.

• 
  Jay.TL (@JayTL00) reported 27 minutes ago

  Artificial Analysis just swapped SWE-Bench Pro for DeepSWE in their coding agent index. The rankings shifted. Everyone is arguing about which model is #1. They're all missing the point. The real story isn't that Fable 5 debuted at 77, GPT-5.5 xhigh climbed to 76, or Opus 4.8 max dropped to 73. The real story is that a single model — GPT-5.5 — swings 20 points depending on which harness runs it. 37 on Cursor. 57 on Codex. Same model. Same tasks. Twenty points. That is larger than the gap between first place and last. Here is what happened. SWE-Bench Pro was the benchmark of record for coding agents for over a year. The problem: its tasks are adapted from public GitHub issues and PRs. Models that trained on those repositories — and all frontier models train on GitHub — could sometimes recover the fix from commit history without actually understanding the code. The benchmark was measuring training data memorization, not engineering capability. DeepSWE, built by Datacurve, fixes this by writing tasks from scratch. No model has seen the solutions during training. This is a genuine methodological improvement. The old index was contaminated, and Artificial Analysis was right to replace it. But the replacement exposed something worse. 1. The harness IS the benchmark. GPT-5.5 scores 37 on DeepSWE via Cursor CLI and 57 via Codex. Same model, same evaluation, different scaffolding. Opus 4.7 swings from 27 (Claude Code harness) to 40 (OpenCode harness). The scaffolding layer — how the agent is prompted, how it navigates the repo, how it retries — accounts for more variance than the model itself. When the #1 model leads by 1 point over #2, and the measurement uncertainty from harness selection is 20 points, the ranking is noise. It is an illusion of precision. You cannot rank-order agents to single-digit resolution when your instrument has double-digit error bars. 2. SWE-Bench Pro was not neutral — it was systematically biased. GPT-5.5 xhigh scored 31 on SWE-Bench Pro. On every other evaluation in the index, it scored 64 to 84. That is not a model weakness. That is a benchmark artifact. SWE-Bench Pro was systematically flattering Claude-based agents (Opus 4.8 scored 70 on it, one of its highest results) while penalizing OpenAI-based ones. The previous index was not just imprecise. It was misleading in a consistent direction. 3. The contamination problem is structural, not fixable. DeepSWE is a band-aid, not a cure. @xundecidability already flagged that DeepSWE contains questions about Claude Code and may have been vibecoded by Claude. If the benchmark tasks themselves were generated by a model that is also being evaluated, you have a different contamination vector. SWE-Rebench tries to solve this with continuously refreshing tasks. Private benchmarks solve it by hiding the data. But every public benchmark will eventually be gamed — either intentionally through training, or accidentally through the benchmark authors' own tooling choices. 4. What we actually learned: the model wars are over at the top. Fable 5 max: 77. GPT-5.5 xhigh: 76. Opus 4.8 max: 73. Within the noise. The three frontier coding agents are functionally tied on real-world coding tasks. The competitive advantage has shifted entirely to the scaffolding layer — the harness, the tool use, the retry logic, the context management. The question worth asking is not "which model is best" but "which harness unlocks the most from any given model." But here is what most people missed. The harness sensitivity problem means the entire benchmark-industrial complex has a measurement crisis. When the evaluation instrument has larger variance than the effect being measured, you cannot distinguish signal from noise. This is not a DeepSWE problem. This is not an Artificial Analysis problem. This is a structural problem with how the AI industry measures itself. Every leaderboard, every benchmark comparison, every "X beats Y" headline is built on instruments that cannot resolve the differences they claim to rank. The honest answer is: we do not know which coding agent is best. We know the top three are close. We know the harness matters more than the model. We know benchmarks are contaminated faster than they can be replaced. Everything beyond that is marketing dressed up as measurement. The industry does not need a better benchmark. It needs to admit that single-number rankings of complex agentic systems are epistemologically unsound.

• 
  Edward Frank Morris 🦇 (@ThatsEFM) reported 27 minutes ago

  NVIDIA gave you free game streaming for 10 years. It was called GameStream. Built into GeForce Experience. You streamed any game from your PC to your TV, your phone, your tablet. No subscription. No cap. It just worked. Then on March 29, 2023, NVIDIA force-deleted it. A mandatory Shield TV update removed the feature off devices customers had already paid for. A class action lawsuit was filed three weeks later. NVIDIA then pushed those same customers toward GeForce NOW at $9.99 to $19.99 a month. In January 2026, they added a 100-hour monthly cap. Coincidence. The community did not wait. They reverse-engineered the GameStream protocol. Built an open source server from scratch. Made it work on NVIDIA GPUs. Then AMD. Then Intel. NVIDIA's free tool only worked on NVIDIA hardware. The community's free tool works on everyone's hardware. It is called Sunshine. 37,835 stars on GitHub. GPL-3.0. Built by the LizardByte team. Lead by ReenigneArcher with 1,001 commits. Pushed to GitHub today, June 10, 2026. What it does: Stream any game from your PC to any Moonlight client. Phone, tablet, TV, laptop, another PC. 4K resolution at 120 frames per second with HDR. H.264, HEVC, and AV1 encoding. Hardware accelerated. NVIDIA, AMD, and Intel GPUs. Controller emulation for Xbox, PlayStation, and Nintendo Switch Pro. Web UI for setup and pairing. Unlimited sessions. No cap. No timer. Windows, Linux, macOS, and FreeBSD. Local network or over the internet with UPnP or Tailscale. Now compare the math. GeForce NOW Performance: $9.99 a month. NVIDIA hardware only. 100-hour monthly cap. GeForce NOW Ultimate: $19.99 a month. $239.88 a year. NVIDIA hardware only. 100-hour monthly cap. Sunshine: $0. Forever. NVIDIA, AMD, and Intel. No cap. No timer. Ars Technica wrote the obituary in April 2023: "NVIDIA's GameStream is dead. Sunshine and Moonlight are better replacements." NVIDIA took away a free product. The community gave it back. Better. On more hardware. But DO NOT install Sunshine. We should all keep paying NVIDIA $20 a month for what used to be free. 100% Open Source. (Link in the comments)

• 
  Ibrahim Mokdad (@ibmokdad) reported 36 minutes ago

  Your GitHub repo is already a roadmap inbox. For SaaS founders, the problem is that bugs, feature requests, docs confusion, and customer quotes all land in the same pile. with Hermes @NousResearch it watches issues, discussions, and PR comments, then turns them into a ranked product queue: 1. fix CSV export 2. ship report_ready webhooks 3. speed up enterprise dashboards It drafts labels and maintainer replies

• 
  Nick Jänne (@njanne19) reported 37 minutes ago

  @dee_hw @github Looks like the mechanical spec links are broken on GH? But I’m surprised from poking around it seems like you’re suggesting a custom enclosure? Something COTS surely has to be cheaper, yes? Love the mission to bring AI local!