Stats
Actions
Tags
From armor
Sketches and tests multiple distinct approaches before committing, with an ego guard to prevent attachment to the first idea. Use at the start of non-trivial, hard-to-reverse tasks.
How this skill is triggered — by the user, by Claude, or both
Slash command
/armor:scenario-spreadThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
The first idea has an unfair advantage. It got there first, and the longer you
The first idea has an unfair advantage. It got there first, and the longer you hold it, the more it looks like the only idea. The fix is to explore several scenarios and let feedback choose. The ego does not get to drive.
Before writing the real thing, sketch three distinct approaches. Distinct means different in shape, not three flavors of the same idea. If three is a stretch, find two.
For each, write three lines: the shape, what it is good at, what it costs.
Get real feedback, not imagined feedback.
mirror-of-self between the top two.The ego guard. Name the approach you walked in wanting. Hold it to the same evidence as the others. If it only wins on attachment, drop it. Feedback that you steer toward your favorite is not feedback.
Pick. Write one line: why this one, and what would change the decision.
For a large task, this is a judge panel. Generate the approaches in parallel, score each against the evidence, then synthesize from the winner while taking the best parts of the runners-up. The Workflow tool runs this shape well.
Trivial or reversible decisions. Spreading three approaches to name a variable is waste. Use this where the choice is expensive to undo — a schema, a public API, a dependency you will live with for years.
npx claudepluginhub markacianfrani/armor --plugin armorGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.