cs-review

只读审查叶子执行器。默认审当前 diff，也审 design 方案，或按用户要求做 repo 级审计。单轮返回结果，不创建子 agent。曾名 cs-code-review。

Quality

68%

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

Fix and improve this skill with Tessl

tessl review fix ./plugins/codestable/skills/cs-review/SKILL.md

Quality

Content

71%Weight 40%Scale 1-5

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 well-structured, actionable read-only review policy with concrete gates, finding levels, and conclusion tiers, plus a follow-up feedback loop. Its main weakness is the dense run-on prose in 调用边界, which hurts conciseness and readability without adding capability.

Suggestions

Break the long run-on sentences in '调用边界' (lines 15-17) into a bulleted checklist so the target-freezing and packet rules are scannable.

Make the single-round review sequence explicit as a short numbered workflow (receive frozen target → diff against intent → grade findings → return terminal report) to surface the implicit checkpoints.

Tighten or split the '审查标准' paragraph into shorter statements so the intent-vs-structure ordering reads as steps rather than one dense block.

Dimension	Reasoning	Score
Conciseness	The body avoids explaining concepts Claude already knows (no definitions of diff/review) and is information-dense, but the '调用边界' section is written as long run-on sentences that could be tightened into list form; mostly efficient but could be tightened, fitting 3 rather than 4 which requires only minor trimming.	3 / 5
Actionability	Provides concrete, specific guidance — three modes, hard gates ('只读', '叶子执行器', return terminal result), per-finding format ('文件:行号'), finding levels (blocking/important/nit) and conclusion tiers (可合/有条件可合/建议先改再合); absence of code is acceptable for an instruction-only skill, with only minor gaps keeping it below 5.	4 / 5
Workflow Clarity	A clear logical sequence is present (freeze target → review against intent → grade findings → return report) and a follow-up feedback loop is specified (逐项报告 resolved/unresolved/new findings); being read-only, the destructive/batch validation cap does not apply, but explicit within-round checkpoints are minor, fitting 4 over 5.	4 / 5
Progressive Disclosure	The skill is under 50 lines, requires no external references (no references/scripts/assets bundles exist), and is organized into clearly labeled sections (调用边界, 模式, 硬门槛, 审查标准, 收尾); per the simple-skills scoring note this earns 5.	5 / 5
	Total	16 / 20 Passed

Description

66%Weight 40%Scale 1-5

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 clearly conveys what the skill does (graded read-only review of diffs, designs, or audits) with natural trigger terms, but omits an explicit 'Use when...' clause, capping its completeness. Distinctiveness is bolstered by strong behavioral constraints (read-only, no sub-agents).

Suggestions

Add an explicit trigger clause, e.g. 'Use when the user asks to review a diff, a design/contract proposal, or to audit part of the repo.'

Drop or de-emphasize internal jargon like '叶子执行器' from the description; move it to the body and keep user-facing trigger language up front.

Consider adding synonyms/extensions users might say ('review this change', '审一下这次改动', 'code review') to broaden natural trigger coverage.

Dimension	Reasoning	Score
Specificity	Names three concrete review modes — '默认审当前 diff', '也审 design 方案', 'repo 级审计' — plus behavioral actions ('单轮返回结果', '不创建子 agent'); several specific actions with minor coverage gaps, fitting the 4 anchor and below 5 which demands more comprehensive action listing.	4 / 5
Completeness	Clearly states the 'what' (reviews diff/design/audit, single-round, no sub-agents) but lacks any explicit 'Use when...' trigger clause — the 'when' is only weakly implied via '默认' and '或按用户要求', so per the judging guideline a missing explicit trigger caps completeness at 3.	3 / 5
Trigger Term Quality	Includes natural user terms 'diff', 'design 方案', and '审计' that a user would actually say when requesting review; the jargon term '叶子执行器' is not a natural trigger, keeping it below 5, but keyword coverage is good.	4 / 5
Distinctiveness Conflict Risk	The read-only, leaf-executor, no-sub-agent niche ('只读', '叶子执行器', '不创建子 agent') is mostly distinct; minor overlap risk remains with the former identity 'cs-code-review' it explicitly references, fitting 4 rather than the 5 anchor of clear niche with minimal conflict.	4 / 5
	Total	15 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: codestable/CodeStable
Path: plugins/codestable/skills/cs-review/SKILL.md
Commit: 19c796d

Reviewed: 4 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.