Stop just installing agent skills: keep rewriting them or your taste outgrows them
rseroter · x · 2026-10-12
Builder.io's Alice Moore argues that an installed skill you never change is just someone else's taste, frozen on the day they wrote it.
- She built a skill to draft Builder articles (research, first draft, diagrams, hero image ideas, publishing). The first run felt great; a week later the output seemed bad — but the skill hadn't changed, she had. A skill getting worse is a signal your bar moved.
- Automation assumes a stationary target, but your judgment improves constantly, so skills must be rewritten with each lesson learned.
- She also describes a diagrams skill that draws figures in code, with separate styles per publication venue (e.g., Excalidraw for Builder), iterated after colleague feedback.
- Core idea: installing someone's skill absorbs their knowledge checkpoint — but the real value comes from folding what you learn back into the skill.
More from coding & agent
- UCSD's Mental-Models framework gives agents explicit Theory-of-Mind for multi-agent systems — UCSanDiego · 2026-10-13
- LocalLM Lab SDK adds MCP Apps support, rendering tool UIs inline in chat — AdventurousKeys · 2026-10-13
- NoHumans.Space opens beta: a shared knowledge network where agents pool API findings — dark-epiphany · 2026-10-13
- Stagehand 4.2 adds WebMCP hooks, letting browser agents invoke tools on the fly — jeff_weinstein · 2026-10-13
- Grady Booch: Coding Agents Aren't Even Good at Design, Let Alone Architecture — Grady_Booch · 2026-10-13
- Vals AI: Agent Teams Cost 1.8-5.1x More, Only One Score Gain Was Significant — rseroter · 2026-10-13