Published on May 06, 2026 at 18:00 CEST (UTC+2)
Show HN: Red Squares – GitHub outages as contributions (668 points by cianmm)
Red Squares – GitHub outages as contributions
This is a playful visualization that turns GitHub's service outages into red squares on the contribution graph, making downtime visible as a pattern of "contributions." It serves as a humorous commentary on GitHub's reliability and the way we track activity. The project highlights how data can be repurposed to reveal unintended insights.
Some kids are bypassing age verification checks with a fake mustache (24 points by jamdesk)
Kids bypassing age verification with a fake mustache
A TechCrunch article reports that children are easily circumventing age verification on adult websites by drawing on a fake mustache, according to a UK-based nonprofit survey. About half of the 1,000 children surveyed said age checks were easy to bypass. The piece criticizes age verification laws that require uploading government IDs, arguing they create privacy risks and databases of personal information.
Vibe coding and agentic engineering are getting closer than I'd like (31 points by e12e)
Vibe coding and agentic engineering are converging
Simon Willison reflects on how his own use of AI coding tools has blurred the line between "vibe coding" (casual, non-reviewed AI-generated code) and "agentic engineering" (responsible, deliberate use of AI). He originally thought these were distinct practices, but now finds them overlapping in his workflow. The post explores how this convergence challenges developers to maintain code quality and oversight.
The bottleneck was never the code (258 points by Anon84)
The bottleneck was never the code
This article argues that while AI coding agents dramatically boost individual productivity, the real limiting factor in software development is human collaboration, not code writing. It cites Fred Brooks and Gerald Weinberg to emphasize that software is a team effort. The author remains skeptical that individual gains will translate to industry-wide speed without addressing coordination and system complexity.
Agents can now create Cloudflare accounts, buy domains, and deploy (516 points by rolph)
Agents can now create Cloudflare accounts, buy domains, and deploy
Cloudflare announces that coding agents can now provision accounts, start paid subscriptions, register domains, and obtain API tokens on behalf of users—all without human intervention beyond initial permission. This is enabled by a new protocol co-designed with Stripe. It represents a major step toward fully autonomous deployment pipelines for AI-driven development.
Setting up a Sun Ray server on OpenIndiana Hipster 2025.10 (85 points by jandeboevrie)
Setting up a Sun Ray server on OpenIndiana Hipster 2025.10
A detailed technical guide walks through installing OpenIndiana Hipster 2025.10 in a Proxmox VM to run a Sun Ray server (a legacy thin-client system). The author shares specific VM configuration steps, including IOMMU settings and disk options. The post caters to hobbyists and retrocomputing enthusiasts interested in reviving Sun Ray infrastructure.
StarFighter 16-Inch (542 points by signa11)
StarFighter 16-Inch laptop
Star Labs launches a premium Linux laptop with Intel Core Ultra or AMD Ryzen 9 processors, up to 64GB LPDDR5X RAM, a 4K 120 Hz matte display, 18-hour battery life, and a removable webcam. It features open firmware options like coreboot and a haptic trackpad. The device targets developers and power users who need high performance with open-source-friendly hardware.
CARA 2.0 – “I Built a Better Robot Dog” (331 points by hakonjdjohnsen)
CARA 2.0 – "I Built a Better Robot Dog"
Aaed Musa presents an upgraded quadruped robot dog using capstan drives, built as a senior design project for under $1,000 and under 20 lbs. The robot is designed for hobbyists and researchers, with a focus on low-cost dynamic actuators. The project includes a build guide and free BOM, aiming to make robotics more accessible.
The Thinking Plant's Man (2025) (34 points by benbreen)
The Thinking Plant's Man (2025)
This historical article profiles Jagadish Chandra Bose, who in 1926 demonstrated plant "heartbeats" via electrical signals and argued plants experience consciousness. Bose used a snapdragon stem connected to a recording apparatus to show responses to sedatives and stimulants. The piece explores the controversy and legacy of his work on plant intelligence.
Reverse-engineering the 1998 Ultima Online demo server (157 points by notsentient)
Reverse-engineering the 1998 Ultima Online demo server
After ten years of work, the author releases a complete reverse-engineering of the 1998 Ultima Online demo server, translating about 5,000 assembly functions into portable C99. The demo contained the full production server code from mid-1998, restricted to a small map. The project provides a rare look at early MMORPG server architecture and serves as a resource for emulator developers.
Autonomous agent deployment is becoming production-ready
Cloudflare's ability for agents to create accounts, buy domains, and deploy without human intervention (article 5) signals a shift from AI-assisted coding to fully autonomous infrastructure provisioning. This drastically reduces friction for deploying AI-generated applications. Implication: Teams must establish robust permission models and monitoring to prevent runaway costs or security breaches as agents gain root-level access.
The convergence of "vibe coding" and "agentic engineering" challenges developer workflows
Simon Willison's observation (article 3) that these two paradigms are blurring indicates that even conscientious developers are slipping into unreviewed AI-generated code. Why it matters: The quality bar for AI output is rising, but so is the risk of hidden bugs and security flaws. Takeaway: Tooling must enforce review gates and provide transparency into what code was AI-generated, even for experienced users.
Individual productivity gains from AI do not solve systemic software bottlenecks
Article 4 argues convincingly that collaboration and system integration remain the primary constraints—long after coding agents become proficient. This echoes the "Mythical Man Month" lesson. Why it matters: Overinvesting in AI coding tools while neglecting team dynamics and architecture will yield diminishing returns. Implication: Organizations should invest equally in AI-assisted communication, design tools, and modular architectures.
Low-cost open-source robotics enabled by AI design tools
CARA 2.0 (article 8) demonstrates how AI-assisted design and simulation can lower the barrier to building functional robots under $1,000. The use of capstan drives and accessible build guides points to a future where hobbyists and small labs can iterate rapidly. Why it matters: Democratizing hardware accelerates embodied AI research and educational applications. Takeaway: Expect more AI-optimized actuator designs and open-source robot platforms that leverage LLM-based design assistance.
Age verification failures highlight weaknesses in AI-based identity systems
The fake-mustache bypass (article 2) is a simple, low-tech attack that defeats computer-vision-based age checks. This underscores the gap between claimed AI capabilities and real-world robustness. Why it matters: As governments mandate biometric and ID-based verification, adversarial attacks will become more common. Implication: AI models for age estimation need adversarial training on edge cases (e.g., makeup, facial hair), and systems should combine multiple signals (behavioral, device fingerprint) rather than rely solely on visual analysis.
Historical perspectives on intelligence inform modern AI debates
The article on Jagadish Chandra Bose (article 9) reminds us that the definition of "intelligence" has always been contested—plants, animals, and now machines. This historical context is valuable as AI researchers debate sentience, consciousness, and agency. Why it matters: Overly narrow definitions of intelligence can lead to anthropocentric benchmarks that miss emergent behaviors. Implication: AI/ML practitioners should look to interdisciplinary history (e.g., plant signaling, animal cognition) when designing evaluation criteria.
Open-source hardware and firmware gain traction for AI workloads
The StarFighter laptop (article 7) with coreboot and a removable webcam reflects growing demand for privacy-respecting, locally controllable hardware suitable for running AI models. Why it matters: As AI usage becomes more personal, users want assurances that their data and models aren't transmitted to cloud servers. Takeaway: Expect more Linux-first laptops with open firmware, dedicated AI accelerators, and hardware kill switches—catering to developers who run local LLMs and need deterministic performance.
Analysis generated by deepseek-reasoner