Skill TRACE Quality Evaluation Workflow

The Problem

Many Skills look usable while still having vague triggers, shallow FAQs, missing references/ material, or reports that polish wording instead of fixing content. Skill TRACE Quality Evaluation Workflow turns that ad hoc review into a concrete self-check task: after creating or modifying a Skill, it checks whether the work meets a delivery-quality bar and avoids the trap where a report looks good but sub-items are still below 5.0.

How It Works

This skill is intended for authors and maintainers of SKILL.md. It treats the Skill content as the object under review, evaluates the TRACE checklist item by item, and reports below-perfect sub-items, their causes, and the direction of the required fix. The key constraint is practical: every sub-item below 5.0 must be addressed by a substantive change in SKILL.md or a references/ file, such as clarifying triggers, boundary conditions, FAQ depth, scoring details, or sample reports, rather than merely rewriting copy.

It supports several common workflows:

• Strict mode: all sub-items must reach 5.0 to pass, useful before delivery.
• Quick mode: only list sub-items below 5.0, useful for rapid diagnosis.
• Focused mode: inspect one dimension, such as FAQ depth or effectiveness.
• Comparison mode: provide two versions of the files and get a delta analysis, useful for checking whether changes actually improve quality.

Boundaries

It fits individual Skill developers, team maintainers, and reviewers who need a consistent evaluation standard. It is not meant for ordinary user conversations. If the edit only fixes typos or formatting without changing usage semantics, a full TRACE pass may be unnecessary. Scoring is executed by AI against fixed criteria, so it is reproducible but not guaranteed to match Skillhub's official evaluation exactly; missing references/ files can cap the score, so the relevant checklist evidence should be added.