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:

• 
  Vitor Zucher | ויטור צוקר 🇧🇷🇮🇱 (@vzucher) reported 4 minutes ago

  github is down my friends . go touch some grass

• 
  masYNYa (@KijAkubovs86334) reported 5 minutes ago

  Someone rebuilt his Grok Bot digital office twice and shipped the second version with 9 roles instead of 6, and the whole thing runs in one evening of setup for the $40 a month a Grok Premium+ subscription costs. The first version is called Grok Bot Office. Six roles: a Chief, plus Research, Writer, Outreach, Ops, and Finance. The Chief receives high-level objectives, splits work across the five specialists, catches everything downstream, and only pings the human for four decisions - send, spend, publish, delete. Everything reversible runs automatically. The setup takes one evening and connects to Telegram for the approval queue. The second version is called Grok Bot Office NY. Nine roles instead of six. Adds Support, Design, and QA to the original team. Same structure: one boss at the top routes every task down and takes finished work back up for signature. Same reversible-runs-automatic, irreversible-waits-for-yes logic. Same Telegram interface. Same evening of setup. The difference is the specialists you can hire. The first version fits a solo operator writing content and running outreach. The second version fits a two-person shop or a solo operator who has customer support volume and needs a QA gate before anything ships. Neither version requires coding. Neither requires a VPS. Neither requires a technical background. The Chief prompt is a plain English job description that says "you receive objectives, delegate them to the specialists below, escalate to the human only when send, spend, publish, or delete is involved." Each specialist prompt is a job description at similar length. What the setup does not do: It does not act on anything irreversible without a human click. It does not have real-world hands to tape boxes, sign contracts, wire money, or make regulated decisions. It does not replace a real employee for jobs that require licenses, judgment under liability, or physical presence. Every claim to the contrary in the last six months of agent content on X has been a marketing shortcut. What it does do: It removes the coordination cost between agents. When Research finishes a brief, it hands to Writer without you copying and pasting. When Writer finishes a draft, it hands to QA without you scheduling a review. The human still owns the last click. The bots own the handoffs between clicks. I built the six-role version for my own work last month. The setup took a Saturday. The most valuable role turned out to be the Chief, not any of the specialists. Before the Chief, I was still routing tasks between the individual bots myself, which was most of the friction. After the Chief, I only touched the four irreversible decisions. Monthly cost of the six-role setup: $40 for the Premium+ subscription, roughly $30 for Zapier for the tool connections, $0 for the prompts themselves. Total under $80 a month, versus what a virtual assistant equivalent would cost at $1,500 to $3,000 a month. The nine-role NY version costs the same. The two extra roles are extra prompt files. Setup time adds maybe an hour. Both patterns are on GitHub in various forms. The prompts are copy-pasteable. The Telegram integration is a two-step tutorial on the Telegram Bots documentation page. None of this is proprietary. The Chief architecture pattern was published in Anthropic's Building Effective Agents post in December 2024 and has been re-implemented across every agent platform since. Someone else already built both templates. The setup only takes an evening if you copy them instead of designing from scratch.

• 
  Simon Lee (@smoon_lee) reported 5 minutes ago

  @iam_edh I was, it broke like 2 hours ago, so I gave up. I was happily building and testing and the BAM, GitHub goes down

• 
  Goozbach.yml (@goozbach) reported 9 minutes ago

  <insert xkcd about swordfights> "Github actions are down again" :sigh:

• 
  Robby Seventeen (@Robby_Seventeen) reported 9 minutes ago

  GitHub actions seems to have so many issues lately. Who is self hosting alternatives and what's specs do you run.

• 
  B (@QuantumTumbler) reported 12 minutes ago

  @XFreeze Anthropic’s problem isn’t caring about safety. It’s treating institutional judgment as a substitute for inspectability. Claude Code has a public GitHub repo, but the CLI source itself still isn’t there. If the safety argument becomes “trust us, we’re the careful ones,” that isn’t open scrutiny. It’s a priesthood with an API.

• 
  electronic Max (@emax) reported 18 minutes ago

  another day another github outage

• 
  rahul agarwal (@rahul67) reported 18 minutes ago

  Github Actions down again?

• 
  Quentiananmen Squarentino (@fgf_fan) reported 19 minutes ago

  @linuxcatgirl Yeah it should be click to diagnose by opening 30 stackoverflow and github issues tabs.

• 
  Kody (@kodykoala) reported 19 minutes ago

  @kentcdodds @wesbos Sorry, the GitHub actions outage slowed me down a bit. Standby!

• 
  Uptimus (@UptimusApp) reported 20 minutes ago

  Aug 26, 2026 at 16:12 UTC: GitHub incident update. Community reports of site performance issues have subsided. Uptimus is currently monitoring stability before resolving the incident.

• 
  Gilmo (@0xgilllee) reported 22 minutes ago

  A deeper look into PONS' mechanism $PONS has been everywhere on my X feed lately. Bags, burns, $PUMP comparisons, people regretting early sells. What caught me was the fee model. It’s easy to track. Pons earns fees, buys $PONS, then burns it. What’s happening? @ponsdotfamily is a token launchpad on Robinhood Chain. People launch and trade tokens directly from their wallets, while Pons earns fees from each trade. Recent numbers: + $PONS ran from around $20M MC to $80m+ + daily fees reached around $140K + 28.79% of supply has been burned + Pons reported $2.85b+ volume from Robinhood Chain’s $25b DEX volume in around one month That’s over 11% of the chain’s DEX volume. Pons is starting to look like one of the main fee plays on Robinhood Chain. Quick recap In mid-August, $PONS traded at around 0.7x annualized revenue, far below $PUMP. Then Uniswap launched its own launchpad and $PONS reportedly dropped around 78% from the top. Pons recovered market share soon after. By late August: + 28.79% supply burned + ~$140k 24H fees + ~56% launchpad market share + ~$76m MC + ~$4m all-time fees That’s when I started paying more attention. The fees were real, and a big part was flowing back into $PONS. What drives $PONS? The model is pretty direct. More launches -> more trades -> more fees -> more $PONS buybacks and burns. Current fee flow: 1. Each trade pays 1% 2. 70% goes to creators 3. 30% goes to Pons 4. 80% of Pons’ share buys and burns $PONS 5. 20% funds the team and operations I mainly watch new launches, trading volume and buybacks. Robinhood Chain gets busier, Pons earns more fees, and more $PONS gets burned. There are claims that all trading fees from $PONS itself are burned. I’d check that part on-chain. The docs are clearer around the 80/20 launchpad fee model. 1) How Pons works today Launch a token and you get: + 1b supply + Uniswap V3 pool with WETH + locked liquidity + trading right away Graduation happens around 4.2 ETH in the pool. The first two blocks also have anti-snipe limits. After that, trading runs normally. 2) What about V2? V2 is being built on GitHub. Tokens will start on a bonding curve before moving to Uniswap V4 with stronger liquidity locks. The current product runs on the V3 model. V2 is the next step. 3) The $PONS loop Token launches
-> Users trade
-> 1% fee
-> 70% to creator
-> 30% to Pons
-> 80% buys $PONS
-> Burn
-> Supply drops Creators get paid in ETH. I like that setup because their revenue stays separate from the token they launched. One detail matters here. The 80% buyback ratio still works as a protocol policy through TWAP. The team plans to make it immutable later. According to Ponsinomics, TWAP buys happen roughly every 15 minutes before the $PONS goes to a burn address. 4) Why did $PONS run so hard? A few drivers: + nearly 29% of supply has already been burned. + more attention brings more launchpad volume, fees, and buybacks. + $PONS started from a low revenue multiple, so strong fee data quickly changed how the market priced it. + RH has around $25b in volume, with Pons taking a decent share. For me, fees and market share matter more than listing rumors. 5) More is building around Pons There are also projects using creator fees for $PONS buybacks, burns, airdrops, raffles, and liquidity. Pons Dividends pays holders in $PONS, while others are exploring lending for $PONS holders. Useful extras, but the main engine remains the fee and buyback model. Where does Pons sit on Robinhood Chain? I see Pons as the main token launchpad on Robinhood Chain right now. $CASHCAT is closer to the meme bet. $PONS is closer to the platform earning fees from the whole game. What I like: + strong market share + recovered after Uniswap entered + low valuation compared with revenue + direct exposure to launchpad activity What I’m watching: + how long current volume lasts + the 80% buyback policy + wash trading + RH activity moving into other sectors @theunipcs mentioned $200m as a possible target, around 10% of $PUMP’s valuation. I can see the case if fees stay strong next quarter. That matters more than any listing rumor. What's my thoughts? I like $PONS because the money flow is easy to follow. Trading creates fees, fees buy $PONS, then supply gets burned. But $PONS already moved from around $20m to $80m MC pretty fast. So I’m watching 7-day fees, burn rate, protocol ETH balance, market share and buyback policy. For me, $PONS comes down to one thing: can Pons keep this level of activity going? If yes, the fee and burn model stays interesting. If volume fades, the whole setup looks very different.

• 
  Chris 👽 (@VeryBullishGuy) reported 24 minutes ago

  @namespacelabs this is one of the smartest capex calls in infra right now and its not even close. apples license only lets macOS run on apple hardware, and only two VMs per machine. thats not a policy on a pdf somewhere, its enforced in code. try to boot a third one and the framework just errors out. so if you want to sell iOS CI you have to physically own the metal. no way around it. thats why AWS has to lock a dedicated host for a 24 hour minimum just to stay compliant, which is completely useless for a 6 minute build. and its why github charges about 10x for a mac minute vs linux, 6 cents against half a cent. that spread IS the business. apple killed the mac pro in march and has no server product at all now. so custom shelves full of consumer hardware is the only play anyone has. this isnt an unboxing video, its a barrier to entry.

• 
  Guzman Pintos (@guzmanpintos) reported 24 minutes ago

  We open-sourced Jardinero, the AI factory that maintains our code at @luxor. - Our log reviewer sweeps our production log 24/7, investigates anomalies and opens pull requests for bugs nobody reported yet - Tag Jardinero on a Linear issue to implement a code change or on GitHub to address comments, fix a flaky test or resolve merge conflicts - Dispatch workflows and tasks directly from Discord Every agent runs in an isolated @tenki sandbox.

• 
  Ayush Singh (@advolt7) reported 26 minutes ago

  follow him if you want to know whether github is down or not, useful.