# Ultrathink Craftsmanship

> Use for expensive, one-pass reasoning gates that stress-test a specification, plan, or near-final implementation for bias, hidden coupling, weak evidence, and craftsmanship risks.

- Skill: `dyai2025/ultrathink-craftsmanship` (Agent Skill)
- Install (CLI): `npx skillmds@latest add dyai2025/ultrathink-craftsmanship`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dyai2025/ultrathink-craftsmanship/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: DYAI2025 (https://skillmd.com/u/dyai2025)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/dyai2025/ultrathink-craftsmanship

---


# Ultrathink Craftsmanship

A deep reasoning gate, not a correctness proof. Run once per configured gate unless the workflow explicitly allows another iteration.

## Modes
- `full`: specification sanity and architecture risk review.
- `kurz` / `kurz+`: compact iteration judgment near release.

## Checklist
- What claim is least evidenced?
- What hidden assumption would invalidate the plan?
- What edge case crosses module, data, security, or UX boundaries?
- Where could tests pass while production wiring is absent?
- What should be removed, simplified, or deferred?

Pair with `konfabulations-audit` for external claims.