Member-only story
Building the Harness Around AI Coding Tools
The basics like permissions & sandboxing, AGENTS.md, rules, docs, memory, planning, subagents & skills
People are building entire workflows and systems around these AI coding tools, often referred to as extending the harness.
Some people make it more confusing than it is, but there is quite a bit to know.
So I thought it was worth covering the basics, like setting permissions and sandboxing, writing the AGENTS.md file, setting up rules & the project’s source of truth, and digging into how memory works.
But also how to build and run subagents & Skills, how to combine them, and how to evaluate them.
I’m also sharing links to bigger projects and deciphering popular bundled Skills and commands like /loop, /goal, and /batch.
This is definitely a how-to (using both Codex & Claude Code). But I’m also doing a bunch of other things like going through the research, comparing approaches, sharing my own failures, and questioning whether some of these things actually improve performance.
