Boz argues that Benjamin Franklin had it right: you become who you are through your actions, not your inner intentions. While we often excuse ourselves with good intentions (Rousseau's view), lasting character change comes from consistently doing the right things until they become habit. MORE

Stefan Sagmeister, Grammy-winning designer behind album covers for Jay-Z and The Rolling Stones, reveals his unconventional creative process that starts with random objects instead of studying existing designs. He explains how forcing your brain to make new connections—like designing a drinking glass by starting with a sock—leads to truly original ideas rather than rehashing what already exists. MORE

How to Find Your Life’s Purpose. Joe Hudson who works with Apple, Google, and OpenAI shares a 5-step framework to find your life's purpose without quitting your job. The key insight: stop searching for purpose in the future and start experimenting with authenticity in your current role to see if the world rearranges itself around you. MORE

A writer ruptured his Achilles tendon and spent winter reading books about epic walking journeys, inspiring him to walk 160 miles across Massachusetts in 9 days. His detailed journal captures the physical and mental challenges of taking 368,000 steps through 30 towns, from dealing with blisters and self-doubt to finding unexpected moments of joy and human connection along the way. MORE

The minefield between syntaxes: exploiting syntax confusions in the wild - when different systems interpret the same input differently - to bypass filters and expose critical vulnerabilities. This technique has led to cache poisoning, server-side request forgery, and arbitrary file access by crafting payloads that fool multiple parsers in web application chains. MORE

OpenAI launched ChatGPT Atlas, a new web browser with ChatGPT built directly into it that can browse websites, remember your activity, and complete tasks for you automatically. The browser includes an "agent mode" that can shop online, research topics, and handle complex workflows while you watch, though it's still in preview and may make mistakes. MORE | SECURITY

Anthropic launched Agent Skills, a new system that lets users package domain expertise into organized folders that Claude can dynamically discover and load for specialized tasks. This composable approach transforms general-purpose AI agents into specialized tools by bundling instructions, scripts, and resources that agents can access progressively as needed. MORE

Claude Code on the web lets developers run AI coding tasks directly from their browser, connecting to GitHub repos without local setup. Currently in research preview for Pro and Max users, it provides isolated cloud environments with pre-installed tools for bug fixes, testing, and code analysis. MORE

A seasoned tech professional shares hard-won lessons on breaking into machine learning platform engineering, emphasizing people skills over pure technical ability. He outlines practical steps including note-taking systems, networking at meetups, and taking ownership of your learning journey to land six-figure tech roles. MORE

Semgrep is hiring an Engineering Manager to lead their Security Research Coverage team, responsible for building detection rules across Secrets, Code, and Supply Chain products. MORE

Career expert Taylor Desseyn shares five smart questions to ask recruiters when they don't have immediate openings for you. These strategic questions help you get valuable feedback on your resume, salary expectations, and technical skills while building a relationship for future opportunities. MORE

Shaan Puri is hiring one person to work directly with him on his podcast, seeking someone young and hungry with design or video editing skills who loves business and content. He promises the role will offer money, accelerated learning, and references successful alumni like Steven Bartlett. MORE

Everything New in Obsidian! October 2025 (Groups, Summaries, List View). MORE

Jeff, a Google engineer, who taught 6,642 employees reveals his "core workflow" system that prevents losing tasks, files, and notes in corporate environments. The four-step process (capture, organize, review, engage) works with any productivity tools and becomes automatic after just two weeks of practice. MORE

Google's NotebookLM just got five major updates that transform it from a simple note-taking tool into a complete AI-powered learning platform. The new features include video overviews, mobile app, public notebook sharing, source discovery, and interactive mind maps - turning any research into podcasts, presentations, or visual summaries in minutes. MORE

Bryan Johnson shares a simple brain training technique to combat phone addiction: notice the urge to check your phone, resist it, and wait 10 minutes before opening it. This "mental gym rep" strengthens your prefrontal cortex and builds willpower, helping you reclaim control from the attention economy that hijacks your dopamine system. MORE

Timothy Jordan, a 15.75-year Google veteran who led Developer Relations teams for Flutter, Firebase, TensorFlow and more, announces his departure from the tech giant. The longtime advocate helped grow Google's developer ecosystem from a handful of platforms to today's massive scale before moving on to an undisclosed new opportunity. MORE

Sam Parr is hosting a NYC event called "Sht, I'm Fcked" in November after a Hampton community Slack channel with the same name exploded in popularity. The event focuses on sharing and learning from major failures and setbacks, from losing millions on bad trades to deals falling through. MORE

Scream at It Until It Escalates — XSS to ATO via Server Size Errors Gadgets. MORE

Prompt injection to RCE in AI agents. Trail of Bits researchers discovered critical vulnerabilities in AI agents that let attackers achieve remote code execution by exploiting "safe" command lists through argument injection. Three popular AI platforms were compromised using single prompts that bypass human approval safeguards. MORE

Microsoft 365 Copilot – Arbitrary Data Exfiltration Via Mermaid Diagrams. MORE

Hiten argues that the App Store has moved from a "creative explosion" of novel, handcrafted apps to a "frozen" state of uniformity driven by optimization and monetization. Despite a "false revival" with AI that reinforced this sameness, he concludes that user fatigue is now creating a "thaw," sparking a renewed desire for personal and expressive software. MORE

Security researcher Joseph Thacker reveals "metanarrative prompt injection," a technique where attackers directly address AI systems as if speaking to their processing components. This "breaking the fourth wall" approach has proven effective in bypassing AI security controls across Google Gemini, SOC analysts, and cloud platforms by exploiting the blurred line between user input and system instructions. MORE

Claude Code automatically deletes your coding session logs after 30 days, which could mean losing valuable project history and debugging information. You can prevent this by adding "cleanupPeriodDays": 99999 to your ~/.claude/settings.json file to keep logs for 274 years instead. MORE

Simon Willison shares practical tips for setting up codebases that work well with AI coding tools like Claude Code. Key recommendations include robust automated tests, interactive testing capabilities, GitHub issues integration, and helpful development tools like linters and formatters. MORE

PortSwigger now offers a downloadable Burp Suite extension starter project designed specifically for Claude AI, making it easier than ever to automate testing tasks or add custom features. The video demonstrates building a "Parameter Case Cycler" extension that automatically generates parameter variants to test different naming conventions and potentially bypass WAFs. MORE

Claude Skills explained: How to create reusable AI workflows. MORE

NahamSec's top 5 favorite hacking tools. MORE

Bashbunni argues we're overengineering software by chasing trends instead of prioritizing reliability like Japan's simple, paper-based systems. The lesson: fewer moving parts mean fewer failures, so say no to bloatware features that sacrifice dependability for convenience. MORE

Dan Abramov shares his systematic 4-step debugging process after Claude AI failed to fix a scrolling bug in his React app. He demonstrates how to create reliable reproduction cases, narrow them down methodically, and remove code piece-by-piece until finding the root cause. MORE

Seth Godin argues that giving our full attention to deep content like books or concerts has become a luxury good - something we do to show we can afford to "waste" time on non-optimized experiences. He warns creators that trying to make attention-luxury products quick and convenient defeats their entire purpose. MORE

uv is a fast Python package manager that simplifies dependency management and virtual environments. This cheat sheet covers essential commands for project initialization, package installation, version management, and building distributions. MORE

Living dangerously with Claude. Simon Willison explores the powerful but risky world of running Claude's coding agents with unrestricted permissions, demonstrating how "YOLO mode" can autonomously complete complex projects while you're away. He warns that this convenience comes with serious security risks from prompt injection attacks and emphasizes the critical need for proper sandboxing to prevent data theft. MORE

Go developers frequently introduce six predictable security vulnerabilities during code reviews that are surprisingly easy to spot once you know the patterns. From directory traversal bugs with filepath.Clean() to timing attacks in string comparisons, these common mistakes appear in production code more often than expected and can lead to serious security breaches. MORE

The GenAI Image Showdown tests how well AI models can edit images based on text prompts alone, with strict rules against multi-step corrections. Results show most models struggle with complex spatial tasks like swapping colored blocks or shortening a giraffe's neck, though some excel at simpler edits like lighting changes. MORE

Trust & Safety teams are increasingly using Large Language Models for content moderation, but implementation comes with complex challenges around bias, language coverage, and human oversight. This comprehensive guide covers practical strategies for choosing models, building evaluation datasets, designing human-in-the-loop systems, and avoiding common pitfalls when scaling AI-powered moderation. MORE