Skills

Change your agent's behaviour with one small edit, not a long rulebook

A change to a skill, rule or hook ships only if it is small, one or two intentful edits, because every added line carries claims nobody checked.

Free, to use and to change. What it can see: only the chat you add it to.

Its short name: skill-design. Source on GitHub.

Install

Pick the one you use.

Claude Code

/plugin marketplace add mbsi-labs/skills
/plugin install skill-design@skills

Changelog

Licence: MIT.

Read the skill itself

Mackenzie (2026-10-16 13:58): "our lever is environment changes, and that pareto-optimal environment changes are small, and they are small because data-vomit sideloads a bunch of implicit claims that cause runaway misalignment."

A change to a skill, rule, CLAUDE.md or hook ships only when all three hold:

  1. Small. One or two intentful changes. Every added line carries implicit claims, so a long change is refused, not trimmed later.
  2. Researched. You looked up the problem the change is for, in sources rather than memory.
  3. Tested. A run with the change and a run without it show the behaviour it was made for.

Write it in a technical register: a defined term ("user-facing view", "hydration") carries what a sentence of plain words would. Name the class, not instances; a list of examples reads as exhaustive, and whatever is missing from it reads as exempt (Mackenzie, 2026-10-28 13:56).

For a skill whose work reaches people (a page, copy, a brief, a persona), have Aion 3.5 (the aion CLI) draft it from the ground up, centred on what people feel, and sample its examples from a domain the work isn't in, so they don't leak; then test it as above. Ask with aion ask.