# We've been using AI wrong: Agent Skills and the Semantic Powers of LLMs

Canonical page: https://dominiklukes.net/writing/weve-been-using-ai-wrong-skills-and-the-trouble-with-context

The guide explains how Agent Skills combine language-model judgement with reusable instructions, files, and scripts to support reliable workflows within context and memory limits.

Blog post · Schemas & Propositions · 2 February 2026

## Where it lives

- [Read it on Schemas & Propositions (schemasandpropositions.substack.com)](https://schemasandpropositions.substack.com/p/weve-been-using-ai-wrong-agent-skills)

## About this item

- Type: Blog post
- Published: 2 February 2026
- Published in: Schemas & Propositions
- Language: English

Topics: ai-agents, context-engineering, tool-use, llm-semantics, llm-memory.

## Related

- [Common misconceptions about Large Language Models and a few literal metaphors](https://dominiklukes.net/writing/what-models-want) (Blog post · 4 August 2026)
- [50 First Dates and the ChatGPT Memory Problem](https://dominiklukes.net/writing/50-first-dates-and-the-chatgpt-memory-problem) (Blog post · 27 August 2025)
- [How Humans Hallucinate and How We Deal with It](https://dominiklukes.net/writing/how-humans-hallucinate-knowledge-memory-and-certainty) (Blog post · 7 December 2025)
- [Karpathy Thinks in Code](https://dominiklukes.net/writing/karpathy-thinks-in-code) (Blog post · 31 August 2025)
- [Symbols and Manipulation](https://dominiklukes.net/writing/symbols-and-manipulation) (Blog post · 8 September 2025)
- [How I (Claude Code) remade this site](https://dominiklukes.net/writing/2026-04-19-how-i-claude-code-remade-this-site) (Blog post · 19 April 2026)

Section: [Writing](https://dominiklukes.net/writing)