Skills

In short: ByteDance’s MUSE-Autoskill treats an agent skill as a lifecycle asset that gets created, remembered, tested, patched, and migrated, instead of a throwaway prompt. On the SkillsBench benchmark, human-written skills lifted it from 53.19% to 68.40%, and on the 35 tasks where it generated its own skill, accuracy reached 87.94%. Your coding agent spends twenty minutes working out a tricky deploy step. It works. The next day you hand it a nearly identical task, and it starts from zero: the same dead ends, the same twenty minutes. It read the docs, ran the commands, and even jotted down a lesson, but that lesson stayed trapped inside one task. When the task ended, the experience went with it. Agents forget, and for anyone who uses them daily this is a familiar, costly habit. On May 27, 2026, the ByteDance team released MUSE-Autoskill to attack exactly that: how an agent turns the experience it builds while doing tasks into skills it can reuse over the long term. There is more than one way to solve continual learning for agents. Some update the model weights, some optimize the outer workflow, and some externalize experience into memory and skills. This article focuses first on the two approaches most closely tied to skills. ...

In short: Microsoft Research’s SkillOpt turns AI agent skills into trainable artifacts. Instead of hand-writing CLAUDE.md, AGENTS.md, or best_skill.md and hoping the rules work, SkillOpt runs the agent, studies its failures, applies bounded text edits, validates the candidate skill, and keeps only changes that improve performance. Every serious AI agent user eventually starts writing instruction files: CLAUDE.md, AGENTS.md, best_skill.md, project rules, tool-use notes, formatting constraints, debugging routines. The pattern is familiar. You watch the agent fail a few times, write a better rule, rerun the task, then add another note. After a while, the instruction file becomes a small operating manual. If you work with Claude Code, Codex, Cursor, or any agent that lives inside a real project, this file quickly becomes part of the product. It tells the agent how to inspect files, when to run tests, how to format answers, which tool calls are safe, what to avoid in production code, and how to recover from common mistakes. The problem is that most of these files are written by feel. You notice a failure, write a rule, and hope the next run behaves better. Sometimes it does. Sometimes the new rule helps one task and harms another. Sometimes the instruction sounds precise to you but remains too vague for the model that has to act on it. ...

Is AI Making Us Give Up Too Soon? What a 1,222-Person Study Revealed In short: A new randomized study (N = 1,222) shows that AI assistance can improve performance in the moment, while reducing independent performance once AI is removed and increasing how often people give up. The strongest negative effect appears in users who ask AI for direct answers. The fix is not to stop using AI, but to change when you bring it in. Ten minutes with an AI assistant. That is all it took, in a new randomized study of 1,222 people, for participants to perform worse on the next problem without AI — and to give up on that problem more often. Not because they were lazy. Because they had stopped expecting hard things to feel hard. This is the second time in a year that careful research has pointed at the same shape of risk. The familiar version of the question is whether AI is making us lazy. Every new tool brings a version of this worry. Calculators made people do less mental arithmetic. Search engines made people remember less. Navigation apps made people worse at finding their way around. ...

On March 31, Anthropic accidentally shipped a source map file in their Claude Code npm package — and for a brief window, the complete TypeScript source (512,000 lines across ~1,900 files) was publicly accessible. The community archived it before Anthropic could pull it down. I spent a few days going through the built-in skills: simplify, batch, skillify, and a dozen others. Most of the community attention went to the hidden feature flags and the easter egg pet system. What caught my eye was less flashy: the way Anthropic’s engineers write their own skills differs from what the official docs teach. Claude Code Skills has two official references — the Skills docs and the Agent Skills Best Practices guide. Both are worth reading. Neither prepares you for what the built-in skills actually look like. This post distills 10 patterns that are in the source but not in the docs. Each one shows a ❌ typical doc-style approach vs ✅ the actual built-in skill approach. If you write SKILL.md files for Claude Code, these patterns change how you structure them. ...

On March 9, 2026, a German indie developer named Cakez77 launched a pixel-art tower defense game called Tangy TD on Steam. He had spent four years building it alone. Within 30 hours, it generated around $30,000. Within a week, it crossed $200,000 in net revenue, with over 28,000 copies sold. During a Twitch stream, he opened his dashboard, saw the numbers, and broke down in tears, hugging his wife. “I can’t believe this many people are willing to support me.” This is often described as “overnight success.” But that framing misses what actually changed. What happened here was structure, not luck. One individual created something valuable. A global platform exposed it instantly. Thousands of people — across countries, time zones, and languages — paid for it at the same moment. For most of history, this wasn’t possible. And the reason it’s becoming common now comes down to one thing: AI is compressing time. For years, a widely accepted idea was be a friend of time — it made sense in a world where time meant accumulation and staying longer meant building deeper advantages. That world is now breaking. ...

I genuinely recommend that you try this AI skill called Last 30 Days, a lightweight AI research tool designed to help you quickly understand what’s happening right now in any topic. At first glance, it looks like nothing more than a small plugin for :contentReference[oaicite:0]{index=0}, but in practice, Last 30 Days works as a real-time trend analysis engine. It scans discussions from the past 30 days across X, Reddit, and the web, then turns that information into structured context that AI can actually use. Whether you’re working on product design, writing cold emails, researching competitors, or simply trying to keep up with the fast-moving world of AI tools, this skill gives you a serious edge. In short, Last 30 Days helps you understand what’s currently happening in any topic—not months ago, not last year, but right now. What Exactly Does “Last 30 Days” Do? The idea is simple but powerful. Last 30 Days automatically searches discussions from the past 30 days across platforms like X, Reddit, and the broader web. It then organizes those conversations into a structured research report that Claude Code can understand and use. ...

Just a year ago, people compared reading lists and book recommendations. This year, nearly every conversation seemed to revolve around AI. There is no denying that AI is an extraordinarily powerful tool. But convenience has a cost. As reliance grows, thinking quietly recedes. As one widely circulated line puts it: “We are trading depth of thought for speed of AI.” A growing body of research suggests this trade-off is real. When MIT Researchers Sound the Alarm Some of the earliest and most serious warnings about AI dependency have come not from skeptics, but from researchers at the forefront of the technology itself. At the MIT Media Lab, research scientist Natalia Kosmina led a striking experiment examining what happens inside the brain when complex cognitive tasks—like writing—are outsourced to AI. Her team recruited 54 undergraduate students from institutions including Harvard, MIT, and Wellesley College. Participants were asked to write SAT-style argumentative essays under three different conditions: Brain-only group: no external tools Search group: access to Google AI group: access to ChatGPT Throughout the task, all participants wore EEG devices to monitor real-time neural activity. ...

The most dangerous mistake knowledge workers can make today is not “not knowing how to use AI.” It’s believing they still have twenty years to adapt. They don’t. We may be the first generation in history forced to watch our core professional abilities overtaken by machines — in the middle of our own careers. It wasn’t like this before. The steam engine replaced muscle. The loom replaced hands. Cars replaced legs. For two hundred years, physical labor was automated. Cognitive labor remained safe. Machines could be powerful, but they didn’t think. That assumption broke in 2023. By 2026, the consequences are becoming impossible to ignore. Those at the frontier are already burning tokens aggressively, leveraging 10x or 50x productivity gains to pull ahead. Most people still haven’t processed what this means. The holidays are a good time to think it through. Acceleration Is the Real Variable Consider adoption timelines. Electricity took 46 years to reach 50% of American households. The telephone took 35 years. Television 22. The internet 7. Smartphones under 5. ...

The biggest misunderstanding about AI skills is that they’re technical. They’re not. They’re behavioral. Tools Change Fast. Work Patterns Don’t. Specific tools will come and go. What lasts is: How you approach problems How you delegate thinking How you evaluate outputs People who benefit most from AI usually change how they work before changing what they use. Skills That Matter More in an AI World AI tends to amplify: Clear thinking Domain understanding Judgment under uncertainty It exposes: Vague communication Shallow expertise Overconfidence Learning AI often feels uncomfortable because it removes familiar excuses. Avoiding the “Learning Trap” Many people learn endlessly and apply little. AI skills become valuable only when they: Replace an old habit Remove friction Save attention If nothing changes in your workflow, nothing compounds. AI doesn’t reward curiosity alone. It rewards integration.