skill-improver

Trigger: improve skills, audit skills, refactor skills, skill quality. Audit and upgrade existing LLM-first skills.

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

87%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a tight, well-structured runtime contract: lean, actionable, with a clean decision table and a verified local reference. The only meaningful gap is that the apply-mode workflow's verification loop is implicit rather than an explicit validate-and-proceed gate.

Suggestions

Make the apply-mode verification explicit, e.g. 'After edits: rerun the audit; only report success when no new issues surface; otherwise fix and re-audit.'

Add a concrete example of the audit output (a short severity-grouped snippet) to the Output Contract so the expected deliverable is unambiguous.

Dimension	Reasoning	Score
Conciseness	The ~45-line body is lean and imperative throughout — table-driven decision gates, terse hard rules, no tutorial background or restating of concepts Claude already knows, within the style guide's token budget. Every line is operational instruction, matching the lean/efficient anchor.	3 / 3
Actionability	Gives concrete, executable guidance: exact file paths (docs/skill-style-guide.md, references/skill-style-guide.md, .atl/skill-registry.md), the exact section list to enforce, a situation-to-action decision table, and a step-3 audit checklist. As an instruction-only skill the absence of code is not penalized because the guidance is specific and actionable.	3 / 3
Workflow Clarity	A clear 5-step sequence with a step-3 checklist is present, but verification is a soft 'rerun or request skill-registry refresh' rather than an explicit validate-then-proceed gate, and apply-mode file edits lack a strong verify-before-finalizing checkpoint. Per the destructive/batch note this caps at 2; not a 3 because checkpoints are implicit rather than bolded validation gates.	2 / 3
Progressive Disclosure	Sections are well-organized in the canonical order and the body points to a real one-level-deep local reference (references/skill-style-guide.md, verified present) that is clearly signaled in the References section. Also qualifies for the top anchor via the under-50-line simple-skill note.	3 / 3
	Total	11 / 12 Passed

Description

85%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is well-formed: it leads with an explicit Trigger clause, states a concrete what, and carves out a distinct niche. Its main weakness is trigger-term breadth — the keywords are natural but repetitive and miss common variations.

Suggestions

Broaden trigger terms beyond the 'X skills' pattern by adding natural variations users would say, e.g. 'review skills, fix skills, normalize skills, update skills'.

Consider adding a brief disambiguator in the description (e.g. 'existing skills, not new skill creation') to further reduce overlap with skill-creator at the discovery layer.

Dimension	Reasoning	Score
Specificity	Names the domain ("existing LLM-first skills") and multiple concrete actions — "Audit and upgrade" plus the trigger verbs improve/audit/refactor — matching the anchor for listing multiple specific concrete actions. Not a 2 because it goes beyond naming only some actions on a domain.	3 / 3
Completeness	Explicitly answers both: an explicit "Trigger: ..." clause for when to use it, and "Audit and upgrade existing LLM-first skills." for what it does. Both what and when are present with an explicit trigger, matching the top anchor.	3 / 3
Trigger Term Quality	Trigger terms "improve skills, audit skills, refactor skills, skill quality" are relevant and natural, but they are repetitive (all built on the word "skills") and miss common variations such as review, fix, normalize, update, or clean up. Not a 3 because the coverage is narrow rather than broadly varied.	2 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (auditing/improving existing LLM-first skills) with distinct triggers, and the body even distinguishes it from skill-creator, making wrong-skill triggering unlikely. Not a 2 because the scope is specific rather than a broad overlap-prone domain.	3 / 3
	Total	11 / 12 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: Gentleman-Programming/gentle-ai
Path: internal/assets/skills/skill-improver/SKILL.md
Commit: e01b114

Reviewed: about 14 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.