the-hat

Evaluates design alternatives using layer splitting, metrics, and tier-list ranking in living notes.

1|Updated Aug 24, 2026
One-click install
npx skills add https://github.com/yamcodes/the-hat --skill the-hat-yamcodes
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: the-hat
Source: https://github.com/yamcodes/the-hat/tree/main/skills/the-hat
Command: npx skills add https://github.com/yamcodes/the-hat --skill the-hat-yamcodes

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Architectural decisions often get made through unstructured chat debates where options are vibe-ranked, orthogonal concerns are forced into false either/or choices, and rejected ideas silently resurface later. This Skill replaces that with a structured 9-step evaluation loop that produces a committed living note in docs/evals/. ## Core Features & Use Cases - Layer Splitting: Decomposes problems into orthogonal dimensions that compose into stacks, avoiding false zero-sum choices between unrelated concerns. - Two-Phase Alignment Gate: Forces the agent to present the problem statement, layer map, option inventory, and metrics first, then pause for user confirmation before scoring. - Tier List with A-Tier Icebox: Ranks options S through E, synthesizes an S-tier stack, and preserves high-value secondary ideas in an icebox without letting them block shipping. - Use Case: When deciding how to wire an API layer, ask the agent to throw the options in the hat; it will split typing vs placement into separate layers, score every candidate against tailored metrics, and commit the tier list to docs/evals/api-wiring.md. ## Quick Start Ask your AI assistant to throw this design decision in the hat and evaluate the alternatives with the the-hat skill.

Frequently Asked Questions about the-hat

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I evaluate multiple design options for an architecture decision?▼

Run the 9-step loop: name the invariant problem, split composing layers, inventory every option in the hat, pick fit metrics, then score each option in prose and produce an S-through-E tier list. The result is committed as a living note in docs/evals/.

What is layer splitting in design evaluation?▼

Layer splitting separates orthogonal dimensions, such as parameter typing versus wrapper placement, so they compose into stacks instead of competing as false choices. Options on different layers are combined, one pick per layer, into a complete S-tier stack.

When should I use a living evaluation note instead of an ADR?▼

Use a living note in docs/evals/ for working decisions that may evolve as options enter or leave the hat. An ADR is only needed when a decision becomes hard to reverse, carries permanent organizational baggage, or requires architectural consensus.

What is the A-tier icebox and when do I ship A-tier items?▼

The A-tier icebox holds highly viable ideas that are not required to close the current change. You do not ship A-tier extras in the immediate PR; they stay documented in the note until a future milestone actually needs them.

Why does the skill stop after defining layers and metrics?▼

The mandatory alignment gate prevents the agent from rushing to a tier list before the problem framing is confirmed. The agent outputs only the problem statement, layer map, hat inventory, and metrics, then asks the user to confirm before scoring.