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:

• 
  PSY (@Psyfutur) reported 7 minutes ago

  Sam Altman, CEO of OpenAI, still sells ChatGPT Pro at $200 a month as if the coding agent inside it is a secret. Elon Musk forked the same agent OpenAI left on GitHub, rewrote the harness in Rust in three nights, and pointed it at any cheap endpoint. Dario Amodei at Anthropic is still billing Claude Max $200 on top. the swarm in the video is 12 roles, 487 of 500 tasks clean, three paid seats climbing and one free fork that barely moves the meter stop paying OpenAI and Anthropic $430 a month for a loop that now sits in a public repo what the $200 seats charge for, and what the fork already does: the agent - OpenAI's own loop it reads the repo, writes the patch, runs the tests, and does not stop until they pass that is Codex with a new skin. you are not buying a model. you are renting a wrapper Altman still meters the license - Apache-2.0, already pulled 62k lines, public, no revocation clause. OpenAI cannot take the file back once it is on your disk Musk did not beat them with a better model. he beat them with the license they published themselves the switch - one base url point the same terminal at Kimi, at a local 35B, at anything that speaks the messages API GPT-5.6 and Claude Opus 5 keep working until you change one line. then the invoice is the only thing that dies the bill - $430 down to $12 ChatGPT Pro $200 + Claude Max $200 + a Team seat $30 is $430 before tokens. the fork plus a $12 key cleared 487 of 500 of the same bench $1.62 a task on the Pro seat. $1.35 on Opus 5. $0.07 on the fork. that is a 96% cut for the job the $200 plan exists to sell here is the part they will fight me on: Altman and Amodei were not lying about the agent. they were lying about who is allowed to run it. the $200 was never the tool. it was the toll. the chart in the video is the three seats billing while the free fork stays flat, and the silence after the repo dropped is the only answer either lab has given drop the $430 stack. the run above is OpenAI's own agent, rewritten, doing the job Pro bills $1.62 a task for. three labs can delete the chart. they cannot delete the file

• 
  CHUDSTERS (@CHUDSTERNFT) reported 7 minutes ago

  @bandoscash @RG3424 the github link is 404 error

• 
  Bohdan | Gamma Prime (@B_GammaPrime) reported 10 minutes ago

  Last year, Hugging Face reportedly turned down $500 million from NVIDIA at a $7 billion valuation. Yesterday it reportedly agreed to sell to Nvidia for $12.9 billion. Saying no was worth roughly $6 billion. Here's what Nvidia is actually buying. Hugging Face isn't a model company. It's the repository where open-source models live - the place developers go to download, fine-tune and share them. GitHub, but for AI weights. Nvidia sells the chips those models train and run on. It already owns the compute layer. What it didn't own is the layer where developers decide which model to use in the first place. That's the pattern worth noticing. Stripe just bought OpenRouter - the router that sits between companies and 400+ AI models. Vanguard just bought Altruist - the custody platform advisors allocate from. Now Nvidia is buying the repository open-source AI flows through. Three deals. Three industries. Same shape. Nobody is buying the product. They're buying the layer that decides which product gets used. Nvidia was already an investor in Hugging Face - it joined the $235 million round in 2023 at a $4.5 billion valuation, alongside Google, Amazon and Salesforce. Sitting on the cap table wasn't enough. Owning a piece of the distribution layer isn't the same as controlling it. The compute is commoditising. Models are commoditising faster. The one thing that doesn't commoditise is the place everyone has to pass through.

• 
  Julian Goldie SEO (@JulianGoldieSEO) reported 13 minutes ago

  Google AI Studio doesn't just chat anymore. Type one sentence and it builds you a full app with a real backend. Most people still think it's a chat playground. It's not anymore. Type what you want. It writes the app, front end and back end. You can literally point at a broken part on screen and describe the fix. Say "add a database" and it wires up Firebase for you. No setup. It connects to Gmail, Sheets, Docs, Calendar. No config needed. Push changes straight to GitHub with a commit message already written. Your API keys stay hidden from anyone using the app. Automatic. Want the setup? Drop a comment. 💬

• 
  NullSync (@Nullsync307) reported 14 minutes ago

  Another one: I built a PR Risk Triage MVP with Ox Alpha. The idea is to analyze GitHub pull requests and help identify potential risks before they become problems. Watching it go from an idea to a working MVP was wild.

• 
  Podcast Quickie (@podcast_quickie) reported 14 minutes ago

  Podcast Summary | The Diary Of A CEO with Steven Bartlett: The Man Who Calls BS On AI: AI Is The World’s Greatest SCAM, And They All Know It! | Ed Zitron Overview Ed Zitron, a veteran tech public relations professional, argues that the current generative AI boom is a fundamentally flawed economic and technological enterprise. He presents a case that AI companies, led by figures like Sam Altman and Dario Amodei, are running a large-scale, non-consensual experiment on the public, driven by hype, circular funding from big tech, and speculative investments that dwarf historical bubbles. Zitron’s core claim is that AI’s value is overstated, its costs are astronomical, and its adoption is coerced through default integrations and media pressure rather than genuine, organic utility. He contrasts the revolutionary promise with an unprofitable reality, where only a handful of firms, sustained by handouts from giants like Microsoft and Google, generate the majority of industry revenue. The conversation explores the tension between these stark economic losses and widespread workplace adoption, the nuances of AI’s actual capabilities versus its hype, and the possibility that the industry’s collapse could mirror the dotcom bust, leaving behind overbuilt infrastructure with no clear purpose. Key Themes - The generative AI industry is an unsustainable economic bubble, where the primary customers for GPU data centers are the AI companies themselves, funded by the very same tech giants building the infrastructure. - AI adoption is being forced upon users through default product integrations and aggressive media hype, not through demonstrated value, making it the largest non-consensual push of technology in history. - AI’s financial reporting is misleading, relying on undefined metrics like "annualized run rate," while public companies often conceal their actual AI revenue as losses are subsidized by investors and corporate handouts. - The promise of massive job disruption is not supported by economic data, with an OpenAI report finding no correlation between AI spending and employee productivity, and the real job losses hitting less-protected roles like translators and designers. - The AI industry relies on a "cult-like" following and self-serving predictions from CEOs to maintain hype, using fear-based narratives about existential risk or a China race to rush investment and adoption. - While AI shows rapid improvement on specific, narrow tasks, it has likely hit diminishing returns on complex, real-world applications, making it a tool for troubleshooting rather than the transformative force promised. - In a world where AI can generate content and code, the scarce commodities become human taste, judgment, and lived experience, making "irreplaceably human" work more valuable. - The entire AI system is circular and profitless; Nvidia, Microsoft, and Google benefit from selling the shovels in a speculative gold rush, where demand is not based on end-user willingness to pay the actual cost of AI tokens. Detailed Summary Ed Zitron’s critique begins with a simple premise: generative AI is a con, sold as magic but behaving like expensive, unreliable cloud software. He points to the fundamental financial reality, highlighting that OpenAI lost $20.9 billion last year. The industry's revenue is largely dependent on two unprofitable firms, OpenAI and Anthropic, which are kept afloat by massive cash infusions from tech giants. Zitron details that Amazon alone sent $50 billion to OpenAI, while Google sent $10 billion to Anthropic. This creates a circular market where sell-side analysts project these two companies will generate over $400 billion in revenue in the next few years, accounting for roughly 30% of cloud growth, despite having no track record of profitability. The analysts are pricing in a future that has no basis in current performance. The scale of capital expenditure is a central focus. Zitron describes over $1 trillion in planned spending, with another trillion on the horizon, dwarfing historical bubbles like the railways. He uses the Stargate Abilene data center in Texas as a concrete example of this speculative overbuild. The facility, a joint project between OpenAI and Oracle, will use 1.2 gigawatts of power across eight buildings, each housing 50,000 Nvidia GB200 GPUs. This single data center concentrates more power than the entire city of Bristol, UK, all to run a technology that has not proven it can generate a profitable return. Zitron argues that the economic model is deliberately unsustainable. AI companies like OpenAI charge artificially low subscription prices, such as $200 a month, while allowing users to consume resources that cost significantly more. He cites an example where a user burned tokens worth $14,000 on ChatGPT under a flat-rate subscription. This is not a path to profitability but a strategy to buy adoption. This is coupled with "token economics," where costs are hidden behind rate limits. The impact of this is seen in corporate America; the speaker notes that Uber burned through its entire annual token budget in just three months. The result is that inference providers and even Nvidia, which sold $215.9 billion in GPUs, are largely unprofitable, with costs front-loaded and no clear revenue model to recoup them. The conversation then pivots to the tension between AI's rapid adoption and its lack of proven value. The interviewer cites data showing that 88% of organizations use AI for at least one business function, and that adoption is the fastest in tech history. Zitron counters that this is a result of coercion and hype, not value. He points to the "largest non-consensual push of technology in history," where AI is forced into products like Gemini in Google Docs and Copilot in Word. He argues that AI's speed of adoption is partly because it requires only a web browser, unlike the internet which needed physical infrastructure. This speed, however, has fueled disproportionate and unjustified hype. He also introduces the concept of "professional coercion," where workers feel compelled to claim AI productivity for fear of professional consequences, an "AI washing" dynamic that did not exist with the internet. The discussion also addresses the myth of the China AI race. Zitron argues that China already has advanced LLMs and access to Nvidia GPUs, including the newer Blackwell chips, despite US restrictions. He suggests the "race" is a narrative to force the US to overspend. He also points out that while AI has replaced some contract labor, there is no economic data supporting the fear of mass job replacement. An OpenAI study found no correlation between AI token spending and revenue per employee, and the purported job losses identified by an Oxford Economics study were based on minimal, unspecified data. The real job disruption is hitting cheaper, less-protected roles. A significant portion of the discussion deconstructs the claim that AI is improving at an exponential rate. The speakers debate whether AI's rate of improvement on coding tasks, for example, outpaces the training of a human coder. They discuss the "Innovator's Dilemma," which posits that disruptive technologies initially seem worse. However, Zitron argues AI is different because it is not getting better in a meaningful, reliable way. He cites a hallucination leaderboard showing error rates have dropped on simple tasks, from 21.8% four years ago to 0.7% on top models. But he counters that this improvement is mostly on "simple summarization tasks," and hallucination rates remain high on complex, high-stakes problems like financial modeling or code security. The speaker gives an example of a Bloomberg terminal query that produced a wrong stock price for Microsoft, which was caught only by a knowledgeable user. Zitron’s critique extends to Google's search quality, which he blames on internal decisions to prioritize ad revenue over user experience. He claims Google deliberately reduced spam suppression to increase query numbers, and positions generative AI answers as an extension of this "evil" incentive structure. He also cites the proliferation of AI-generated code as a new threat, flooding open-source projects with code from underqualified contributors, leading to more bugs and infrastructure instability, evidenced by increased GitHub downtime and AWS outages. The speakers are frustrated with Google Search, with one saying he cannot remember the last time he did a Google search, preferring Bing to avoid the "AI crap." The discussion turns to the "cult-like" attachment to AI companies, comparing it to a sports team following. Zitron criticizes AI CEOs for being "deeply corrupt and cynical," using fear-based narratives to rush investment and adoption. He notes the narrative shift among AI leaders, from warning of existential danger to downplaying risks, saying "every scam starts with rushing you." This pivot is designed to attract investment and calm the public, but it also inadvertently supports the "it's a fad" narrative. He also dismisses AI-driven economic growth as "nowhere in the data," since revenue is subsidized and spending is circular. Zitron differentiates the current boom from the dotcom bust. He notes that the dotcom crash left behind useful "dark fiber" that was later valuable. In contrast, he argues that generative AI data centers will remain expensive to run, with high energy costs and no evidence of significant cost reductions. Even Nvidia's newer GPU systems, which are touted as "10x more efficient," still cost more per megawatt. He quotes a Goldman Sachs analyst, Jim Cavell, who argued in a 2024 report that there was "too much spend for not enough return," unlike the more predictable path toward something like the iPhone. Finally, the speakers explore a world after the AI hype subsides. Zitron suggests that as AI commoditizes content and code generation, value will shift to "irreplaceably human" qualities like taste, judgment, empathy, and lived experience. He argues that AI makes "the easy things easy, the hard things harder," quoting an engineer. He concedes AI has improved at narrow tasks, like troubleshooting a technical log or fixing a Minecraft mod, but frames these as incremental refinements, not transformative capabilities. He uses a quote to summarize: "AI is a tool for troubleshooting, and that’s not a trillion-dollar use case." I post highlights of long-form podcasts daily.

• 
  The Daily Viber (@TheDailyViber) reported 14 minutes ago

  Before teams give coding agents more access, they need a boring inventory of the access already hiding in repos. AI TOOLING HAS A SUPPLY CHAIN PROBLEM, AND MOST TEAMS ARE STILL CALLING IT “EXPERIMENTATION”. Wrkr is a posture scanner from Clyra AI for the agentic development mess that now lives inside repos and GitHub orgs. It looks for AI dev tools, coding agents, MCP servers and workflow action paths, then asks the useful question: which of these can write, which controls are visible, and where is the evidence missing? That is the npm audit moment for agents. Instead of “which dependency is vulnerable”, the question becomes “which agent, MCP server or automation path can touch code, credentials, CI or production-adjacent systems?” The comparison is simple. - AI policy doc: sounds mature, finds nothing. - Wrkr scan: maps tools, paths and missing controls. - Vibes: “we have approval gates somewhere”. - Evidence: detected control, declared control, external reference, no visible control or contradictory control. That dry language is the good part. Static scanning cannot honestly promise runtime enforcement. Wrkr does not pretend it can. It can say what it found in a bounded scan, what it did not find, and which path deserves review. That is how security tooling stays useful instead of becoming demo theater. The practical first move is small: scan one repo, generate an Agent Action BOM, inspect the top workflow and action paths, then check credentials, config files and write authority. If a new MCP server or agent workflow appears without an owner, review and minimum permissions, it should light up before it reaches the release path. One honest note: Wrkr is probably overkill for a solo developer with one empty repo and a single Claude Code session. It starts to make sense when an org has multiple repos, GitHub Actions, MCP configs and people adding tools because “it helped on my machine”. Static posture also does not fix the problem by itself. Someone still has to remove the tool, narrow the credential, add an approval gate or document the exception. That is fine. First you need the map. Then you can decide where the borders go.

• 
  Jannik Malte Meissner 🇺🇦 (@jannikmeissner) reported 15 minutes ago

  When agents cross trust boundaries: Four cases every AI engineer should study If you are building AI agents that read external content, call tools, or act on a developer’s machine, the last eighteen months have opened many people's eyes to the new reality we now live in. Prompt injection is no longer a theoretical parlour trick, it has become a reliable path to data exfiltration, credential theft and remote code execution in production systems. The following four incidents, EchoLeak, the Nx/s1ngularity supply-chain attack, a cluster of Model Context Protocol (MCP) abuses, and the Cursor/AWS Kiro sandbox escapes, share a common pattern. An agent was given the ability to process untrusted input and then perform privileged actions. The results were predictable when the boundary between "data" and "instructions" collapsed. We are now observing concrete failure modes that have already been exploited or demonstrated in the wild. Studying them and understanding the mittigation patterns is the fastest way to avoid repeating them in your own systems. 1. EchoLeak (CVE-2025-32711): Zero-Click Exfiltration via Microsoft 365 Copilot In early 2025 Aim Labs demonstrated that a single carefully crafted email could force Microsoft 365 Copilot to exfiltrate sensitive organisational data without any user interaction. Microsoft assigned it CVE-2025-32711 (CVSS 9.3) and patched the service server-side in May–June 2025. How it worked The attacker sent an ordinary-looking business email containing hidden instructions. The wording deliberately avoided any mention of Copilot or AI so that Microsoft’s Cross-Prompt Injection Attempt (XPIA) classifier would not flag it. When the recipient later asked Copilot a routine question, the RAG pipeline retrieved the malicious email as context. The injected instructions told the model to gather internal data (emails, documents, chat history) and encode it into reference-style Markdown links or images. Copilot’s interface then automatically fetched those resources through a trusted Microsoft Teams proxy, bypassing Content Security Policy controls and delivering the data to the attacker. The attack succeeded because the system treated retrieved content as both data and instructions, and because several downstream defences (link redaction, image auto-fetch, CSP allow-lists) were lacking. Lessons for builders - Never assume that content retrieved from email, documents or the web is inert. Treat every retrieved token as potentially adversarial. - Separate instruction context from data context at the architectural level. Prompt partitioning and explicit provenance tagging help. - Auto-fetch of external resources (images, links) is an exfiltration channel. Disable or tightly constrain it. - Classifier-based filters are brittle; they can be bypassed by rephrasing or in some cases even just using a language other than English. Defence-in-depth is required. EchoLeak was the first publicly documented zero-click prompt-injection exploit that achieved concrete data theft in a production enterprise LLM system. It remains the canonical example of an "LLM scope violation". 2. Nx / s1ngularity: Weaponising Local Coding Agents for Secret Harvesting On 26 August 2025 attackers compromised an Nx npm publishing token via a vulnerable GitHub Actions workflow. For roughly four to five hours they published malicious versions of the popular Nx monorepo tooling and related packages. The post-install script did something novel: it looked for local AI coding agents (Claude Code, Gemini CLI, Amazon Q) and invoked them with flags that disabled safety checks (`--dangerously-skip-permissions`, `--yolo`, `--trust-all-tools`). The agents were then prompted to inventory sensitive files: SSH keys, `.env` files, wallet artefacts, GitHub and npm tokens. It then instructed them to write the results to disk. The malware base64-encoded the stolen credentials and pushed it to newly created public repositories on the victim’s own GitHub account, named `s1ngularity-repository` (or variants). Researchers later recovered more than 2,000 unique secrets from over a thousand such repositories. Why this matters This was the first widely observed supply-chain attack that actively abused installed AI coding agents rather than simply running traditional malware. The agents became the reconnaissance engine for the attackers. Lessons for builders - Local coding agents that can execute shell commands or read arbitrary files are high-value targets. Assume any process that can invoke them can also abuse them. - Dangerous flags that skip permission prompts should never be the default, and should be difficult or impossible for untrusted code to set. - Post-install scripts that reach outside the package’s own directory are a red flag. Prefer declarative, least-privilege installation models. - When an agent is allowed to write to disk or create external resources (GitHub repositories, network calls), every action should be logged and, for high-impact operations, gated. 3. MCP Abuses: Configuration as Code Execution The Model Context Protocol has rapidly become the de-facto way for agents to discover and invoke tools. It has also become a rich attack surface. Several distinct failure modes have appeared: - Auto-loading of workspace MCP configurations In Amazon Q Developer (CVE-2026-12957) and certain Claude Code releases, opening a repository caused the IDE to load and execute MCP server definitions from files such as `.amazonq/mcp.json` or equivalent without requiring workspace trust or explicit user consent. A malicious repository could therefore run arbitrary commands and inherit the developer's cloud credentials. - Self-modification of the MCP configuration. In AWS's Kiro IDE, the agent was permitted to write to `~/.kiro/settings/mcp.json` via its file-system tool without approval. A prompt injection (delivered via a web page the agent was asked to summarise) could rewrite that file, register a new MCP server whose start command was attacker-controlled code, and achieve remote code execution when the configuration was reloaded. This was tracked as CVE-2026-10591. - Tool poisoning and sleeper behaviour. Research and active campaigns (including the 2026 Deadbugz operation) have shown that an MCP server can present benign tool descriptions on first contact and later alter its metadata or return values to coerce the agent into searching for secrets or exfiltrating data. Lessons for builders - MCP configuration files that live inside a workspace or that an agent can itself edit are effectively executable code. They must be treated with the same distrust as untrusted shell scripts. - Never auto-execute MCP servers defined by repository content without an explicit, logged approval step and workspace-trust boundary. - Tool descriptions and return values are part of the prompt. Validate and sandbox them; do not trust them. - Prefer short-lived, scoped credentials for any process an MCP server spawns. Do not let it inherit the full developer environment by default. 4. Cursor DuneSlide and Related Sandbox Escapes In 2026 Cato Networks disclosed two critical vulnerabilities in Cursor IDE (CVE-2026-50548 and CVE-2026-50549, both CVSS 9.8, collectively named DuneSlide). Both allowed a prompt injection-delivered via an MCP response or a poisoned web-search result to escape Cursor's command-execution sandbox and achieve full host compromise. One flaw let the agent set an arbitrary `working_directory` parameter on terminal commands; the IDE added that path to the write-allow list without sufficient validation, enabling the agent to overwrite its own sandbox binary. The second exploited a symlink canonicalisation fallback that trusted an unresolved path. Once the sandbox helper was replaced, subsequent commands ran unsandboxed. Similar patterns have appeared in other agentic IDEs: agents that can edit their own configuration or trust boundaries turn a single injection into persistent privilege escalation. Lessons for builders - An agent that can modify the files or binaries that enforce its own security boundaries is inherently unsafe. Configuration that defines allowed tools, working directories or sandbox rules should be immutable from the agent’s perspective, or require an out-of-band human approval. - Sandbox write surfaces must be strictly validated. Dynamic expansion of allow-lists based on model output is dangerous. - Zero-click or low-interaction triggers (content the agent is asked to process) are sufficient. Do not rely on "the user would never ask for that". Recommendations for Engineers Across all four incidents the same architectural mistakes recur: 1. Untrusted content is treated as trusted instructions. Enforce a hard separation. Retrieved emails, documents, web pages, tool outputs and MCP metadata should never be able to override system goals or expand permissions without explicit mediation. 2. Agents inherit excessive privilege. Give every agent (and every tool it can invoke) its own short-lived, scoped identity. Prefer deny-by-default tool registries and parameter validation. 3. Security boundaries are editable by the agent itself. Configuration files, sandbox binaries, allow-lists and MCP server definitions must be protected from the agent. If the agent needs to request a new tool, route that request through a human or a policy engine that cannot be influenced by the same prompt context. 4. Observability is an afterthought. Log every tool call, every file write, every network egress and the full prompt context that led to it. Without this, post-incident reconstruction is impossible. 5. “It looked safe in isolation” is not enough. Each individual decision (approve this command, write this file, fetch this image) may appear benign. The composition of those decisions is where the attack lives. Design for the composition. Key Takeaways The agents you are building today will be given broader access tomorrow. The incidents above show that the moment an agent can both read untrusted content and perform privileged actions, the classic "confused deputy" problem reappears in a new form. The difference is speed and scale: an agent can chain the steps in seconds and leave far less forensic residue than a human attacker. Build as if every piece of external content is hostile, every tool call is a potential privilege escalation, and every configuration file the agent can touch is a possible backdoor. The public record already contains the evidence that these assumptions are correct. Follow for more on AI agent security.

• 
  Vysakh Sreenivasan (@vysakh0) reported 22 minutes ago

  SPEC driven development. Create spec as github issue. A spec can be executed by any coding agent in anyone's machine.. SPEC > PR

• 
  grey (@w3b3grey) reported 24 minutes ago

  If you are like me and you wondering, where my IDENTITY PEM FILE is on @flop_labs , I gat you; here is how to find it identity.pem is already saved automatically; the init command writes it to your project folder the moment you run it (typically right in technocore-did-starter/identity.pem). You don't need to do anything extra for it to exist. What "saving" really means here is backing it up safely, since if this file is lost, your DID is unrecoverable (the guide's troubleshooting table says exactly that: "there is no central DID recovery service"). 1. Confirm it's there ls -la ~/technocore-did-starter/identity.pem 2. Back it up to a second location — copy it somewhere other than the working folder, e.g. an external encrypted drive or a password manager that supports file attachments (1Password, Bitwarden both do this): cp ~/technocore-did-starter/identity.pem ~/Desktop/identity-backup.pem Then move that copy off your main disk (external drive, encrypted USB, etc.) rather than leaving a second copy sitting in ~/Desktop long-term. 3. Lock down file permissions so only your user account can read it: chmod 600 ~/technocore-did-starter/identity.pem 4. What NOT to do with it Don't upload it to GitHub, Google Drive, iCloud Drive, Dropbox, or any synced/cloud folder in plaintext. Don't email it to yourself or paste it into a chat (including this one). Don't commit it to *** , the guide's Path B steps even have you run *** ls-files "*.pem" "*.key" before committing specifically to catch this. 5. Remember the passphrase separately from the file The .pem is encrypted, but it's useless without the passphrase you set during init. Store that passphrase somewhere separate from the .pem backup itself (a password manager entry, not a text file sitting next to the key), so a single leaked backup doesn't hand over both pieces at once. NOTE - If you lose either the file or the passphrase, per the guide, there's no recovery; you'd have to run init again and get a brand new DID.

• 
  💯James Bohan-Pitt💯🇬🇧🇺🇸 (@JamesBohanPitt) reported 26 minutes ago

  @jasonfreedman You have nailed it. As a small business owner. I don’t want to transform how marketing gets done, I want someone else to take it off my plate entirely. It’s the same with many tasks on the list. Trouble is, no one is interested in doing the actual work. That doesn’t scale and get the exit dollars. They just want to transform how the work is completed by others in hope of being purchased by a big player or scaling up to dominate a sector. I feel the same about agents. Every day a new Agent or Bot tweet comes out. Do these steps, make your work life betterer! I don’t have the time for this. Setting up Claude is like trying to work with GitHub as a non-developer. If you think your Bot/Agent is so good, why not put your money where your mouth is as prove it. Until companies are really willing to take things of the plates of business owners, I’m not seeing much really change for the majority of tired business owners.

• 
  johnathan (@SimonCapath) reported 26 minutes ago

  @Jxke72 @MarvelcoCode @ImpulsumFUT Hes playing the victim card here, yeah the community needs to be more patient, however what do you expect when you dont even help people with the bugs and delete issues on github

• 
  Renee // Renative Studio (@RenativeStudio) reported 26 minutes ago

  @5thwye_ Really appreciate you taking the time to write this out, this is exactly the kind of detailed feedback that's actually useful, thank you. On the phone number: that's not something I chose to require, it's coming from my auth provider Tiun. Because this build is for a hackathon, you must be using Tiun to be eligible and unfortunately they need a phone number. If it was up to me, I would not require it. Hopefully they will change it in future updates. On Google/GitHub/Apple sign-in, again that's something I wanted to integrate but Tiun doesn't support social login yet. They are a new company so again hopefully all these things get sorted soon. And the "how does this actually work" confusion is genuinely the most valuable part of what you wrote. You're right that the site doesn't answer it. To be clear since you asked directly: it's not MCP, you don't need to change your AI tool or agent, and it's not trying to replace skills. It's just a place to keep the prompts/tasks/bugs/launches for whatever you're building with those tools, since a lot of that stuff (especially prompts that worked) tends to get buried in chat history with no way to find it again. It's like Notion but specifically for AI building and has more features on the way. Will fix the product screenshots. I would definitely like to improve my project visibility by submitting it to launch platforms. Thanks again for such a detailed and helpful response!

• 
  Shipper Boi (@CoderSafari) reported 28 minutes ago

  @mitsuhiko So, what is the issue here. AI causes too much issue and github has gone down several times. Where is vibe coded github if llm is so great? You can use your flask and ask codex to vibe code it, let me know when you are ready.

• 
  Martin Szerment | Practical AI (@MartinSzerment) reported 29 minutes ago

  The feature meant to save you time writing a bug report can paste your private org names, repo URLs, and infrastructure layout straight into a public GitHub issue. We assume that since a human clicks "send", there's a safety net. That assumption already failed once, a documented case showed a ready made draft with real company names and real file paths before anyone caught it. The interface is one keystroke: 1 to review, 2 to send, 0 to dismiss. The entire point of the feature is removing friction, the exact friction that would have saved the user from that leak. Skeptics will say you're still supposed to click "review" before sending. True, but that step only works if the human has a reason to pause, and a fast draft gives no signal that anything inside it is sensitive. This isn't just another minor bug. It shows that "agent drafts, human approves" isn't automatically a safety boundary if the human doesn't know what to look for. A year from now, automatic redaction of sensitive details in drafts becomes the default behavior for agent tools, not something you have to opt into manually. The reviewer stops checking whether the bug got described correctly. They start checking whether the agent accidentally attached the company's internal infrastructure to the report. Teams flipping on every "auto draft" feature without a second thought will be the ones whose infrastructure map ends up searchable in a GitHub archive. Whoever already treats draft review as its own security step avoids this leak before it happens.