Topic
Nate B. Jones on Context Engineering
In February 2026, Nate B. Jones used the common definition of context engineering: the set of strategies for curating and maintaining the optimal set of tokens during an LLM task. As of August 2026, he argues that on long agent runs context has to change as the work teaches you what the job is, so he keeps a short current-state file the agent reads and updates.
Start here
Every page on Context Engineering
- Context engineering for long agent runs: CLAUDE.md, AGENTS.md and a current-state file August 12, 2026
- Paste This Into Claude, Never Hit a Token Limit Again July 29, 2026
- Karpathy's Wiki vs. Open Brain. One Fails When You Need It Most. April 22, 2026
- AI Automation: The $0.10 No-Code System That Replaced My SaaS March 2, 2026
- What is prompt engineering? Knowing how to structure a query February 27, 2026
More from Nate B. Jones
Questions
Is prompt engineering the same as context engineering?
Nate treats them as related layers. In February 2026 he described prompting in 2026 as a stack running from prompt craft through context engineering to intent and specification engineering. In August 2026 he said a strong opening prompt still matters and can only express what you know before the work starts. Context has to keep changing as the work teaches you what the job is.
What are the four pillars of context engineering?
In August 2026 he separated four kinds of context for long agent runs: stable instructions, the current project state, a map of what material exists and where it lives, and history. He keeps history apart so it can't masquerade as current instructions.
Last reviewed