Nadeem Shabir 💫

This guy literally broke down how to use Claude Code like an expert: 1:40 - Code vs Cowork vs OpenClaw 6:51 - Setting up context status line 12:03 - Sub-agents 17:49 - Creating skills 23:58 - Ask user questions tool 33:33 - Tool-powered skills: Tavily 36:57 - CLI vs MCP vs API hierarchy 39:30 - Make slides skill w/ Puppeteer 43:32 - Auto-invoking skills with hooks 46:49 - Jupyter notebooks for data trust 55:09 - The operating system file structure

This is disgraceful on so many levels. The Government promised constructive dialogue with Labour MPs opposed to the attack on the right to jury trial, instead, we have this sort of nonsense and press briefings. Utterly pathetic.

Karl Turner MP has had the Labour whip removed. I think that makes Starmer something of a record breaker with 16 + suspensions when it comes to intra-party authoritarianism. Tony Blair didn’t even get close with his four suspensions over 10 years. 1/2

🚨 CRITICAL: Active supply chain attack on axios -- one of npm's most depended-on packages. The latest axios@1.14.1 now pulls in plain-crypto-js@4.2.1, a package that did not exist before today. This is a live compromise. This is textbook supply chain installer malware. axios has 100M+ weekly downloads. Every npm install pulling the latest version is potentially compromised right now. Socket AI analysis confirms this is malware. plain-crypto-js is an obfuscated dropper/loader that: • Deobfuscates embedded payloads and operational strings at runtime • Dynamically loads fs, os, and execSync to evade static analysis • Executes decoded shell commands • Stages and copies payload files into OS temp and Windows ProgramData directories • Deletes and renames artifacts post-execution to destroy forensic evidence If you use axios, pin your version immediately and audit your lockfiles. Do not upgrade.

Most organisations learn backward: commit first, discover problems painfully. However, sleeping rats teach a better way. They "twitch", making small, safe movements to learn their bodies before performing. Organisations should do the same with product launches, reorganisations, and strategy: run small pilots, test assumptions in constrained environments, and build understanding before betting everything. Twitching increases learning velocity without increasing existential risk. Leaders who embrace this posture, treating strategy as exploration rather than declaration, adapt faster and earn more trust. mikefisher.substack.com/p/twitching-be…

In the age of AI, knowing when to slow down is counterintuitive but critical. AI excels at fast execution but amplifies the impact of flawed decisions, the cost of getting requirements wrong goes up precisely because System 1 (execution) has become so powerful. The skill isn't choosing speed OR slowness, but knowing which phase you're in. Spend 10 minutes clarifying requirements, run a pre-mortem with AI, build a throwaway prototype. Teams shipping fastest long-term are often those that slow down at the right moments. theengineeringmanager.com/growth/slow-do…

This little illuminated dragon is very happy about Pretext. He's too busy having fun to care about people's "hot takes" on how "it's not that special." (This little dragon also only works on desktop right now but maybe I'll do mobile later) illustrated-manuscript.vercel.app

My dear front-end developers (and anyone who’s interested in the future of interfaces): I have crawled through depths of hell to bring you, for the foreseeable years, one of the more important foundational pieces of UI engineering (if not in implementation then certainly at least in concept): Fast, accurate and comprehensive userland text measurement algorithm in pure TypeScript, usable for laying out entire web pages without CSS, bypassing DOM measurements and reflow

Child! Not just teenager. Child! And who starved him Sky? Who starved him?

'He's playing fast and loose with the truth.' @andytwelves calls statistics used in @GoodwinMJ's latest book into question, particularly those on English as an Additional Language for young people.

Goodwin gets murdered on live TV

Open source models are approaching parity with frontier labs, raising questions about whether those labs’ valuations are justified if model capability becomes commoditized. While frontier labs point to enterprise features, safety, and distribution as moats, the key metric is not raw capability but the “monetizable spread”, the portion of capability advantage customers will actually pay for. Since most revenue comes from tasks that are already “good enough” with open models, this monetisable spread is shrinking faster than the capability gap. IMHO developments like Sora illustrate this dynamic: even when frontier labs achieve striking capability breakthroughs, those advances don’t necessarily translate into scalable revenue, suggesting the economic moat may be narrower than it appears. This is well worth reading: open.substack.com/pub/davefriedm…

I agree with this piece that Agency, the ability to move from identifying a user problem to shipping a solution without waiting for someone else to direct the next step, matters more than seniority for product engineers. AI tools are compressing the depth work that used to require years of experience. What matters now is breadth: can you talk to users, sketch a solution, build a working version, and ship it fast? The most high-agency people own the full loop and drive ideas to production with minimal oversight. For managers: every approval gate and alignment meeting slows them down. If it takes weeks for an idea to reach users, something is broken. It should take hours. anantjain.xyz/posts/agency

LLMs are great at writing individual functions but struggle with whole projects. Ossature fixes this by combining specs, narrow task context, and verification loops. You write specifications describing what your software should do, Ossature validates them, audits for gaps, and generates code one task at a time with built-in verification. Each task only sees the context it needs and not a 128K token dump. If generation fails, a repair agent fixes it. The result: reliable, coherent multi-component systems built from specs, not guesswork. MIT-licensed and ready to try. ossature.dev/blog/introduci…

Autonomous coding agents fail in predictable ways: they lose focus over long tasks and are terrible judges of their own work. Anthropic's solution: a three-agent harness inspired by adversarial networks. - A planner expands simple prompts into specs. - A generator builds the app. - A separate evaluator actively tests it using Playwright, catching bugs and design flaws. Results are dramatic: a solo agent spent 20 minutes and $9 building a broken game maker; the harness spent 6 hours and $200 to build one that actually worked. The difference: game engine functionality, UI polish, feature completeness were all vastly superior. As Claude models improve (Opus 4.6), the harness gets simpler: remove unnecessary scaffolding, keep only pieces that add value. Core lesson: separate generation from evaluation, provide concrete criteria, and re-examine your harness as models improve. anthropic.com/engineering/ha…

Struggling with an overly confident, combative colleague? Stop fighting them head-on. Instead, use their momentum against them. Build on their ideas positively, reframe objective debates as subjective possibilities, and let their dominance work in your favor. Yue shares two tactical strategies: Sidestep and Redirect, and Reframe and Add, that shift defensive conversations into collaborative ones. The secret: consistency, practice, and keeping frustration out of your tone. news.theuncommonexecutive.com/p/how-to-handl…

The best managers know when NOT to step in. When you do intervene, distinguish between fatal "below the waterline" mistakes where guidance is essential and stylistic differences where you should defer. I do think it's important to use task-relevant maturity, not seniority, to calibrate involvement. Every intervention should be a learning moment, be principled about your reasoning, provide missing context, and leave your team empowered to handle it next time. The goal isn't to make every decision; it's to build a self-sufficient team that understands your vision. dein.fr/posts/2026-03-…

Anonymous announces it will soon release the FULL Epstein files — exposing the pedophile psychopath in the White House and his wealthy child rapist buddies. 😳👇 x.com/TMT_arabic/sta…

Issue tracking systems were built to manage handoffs between PMs and engineers, but that's no longer how software development works. Linear is declaring that era dead and repositioning itself as a context-to-execution platform for human-agent collaboration. With AI agents now authoring a quarter of new issues and adoption growing 5x, the focus shifts from orchestrating process to feeding rich context that agents and humans can work from together. Linear is launching its Agent interface, Skills, and Automations, with Code Intelligence and a native Coding Agent coming soon. The next system isn't designed around handoffs; it's designed around context. linear.app/next