nsfc-humanization

去除 NSFC 标书中的 AI 机器味，覆盖词语、句法、段落和章节层，尤其处理伪对立、工程协议腔、规格书式字段串、术语漂移、边界声明过重和研究动作不清（不适用：非标书内容/需修改格式/需补充新内容）

Quality

70%

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 ./skills/nsfc-humanization/SKILL.md

Quality

Content

78%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 content is lean, actionable, and well-structured with a clear multi-step workflow, explicit validation/feedback loops, and well-signaled one-level-deep references to real bundle files. The main weakness is mild structural redundancy from two overlapping numbered lists that could be consolidated.

Suggestions

Consolidate the two numbered lists under 执行步骤 (the 1-4 layer list and the 1-7 procedure list) into a single unified sequence to remove structural redundancy.

Add one short before/after example showing how a 伪对立 or 工程协议腔 sentence is rewritten, to lift actionability from procedural to demonstrably executable.

Consider moving the long canonical 公共硬约束 block to a shared reference if it is identical across skills, keeping the SKILL.md body focused on this skill's specifics.

Dimension	Reasoning	Score
Conciseness	The body is dense and assumes Claude's competence — it does not explain what NSFC, AI-flavor, or basic editing concepts are — but the two overlapping numbered lists (1-4 layers then 1-7 steps) and the canonical common-constraints block could be trimmed or reorganized slightly, keeping it just below level 5.	4 / 5
Actionability	Concrete procedural guidance is present: a four-layer scan, explicit decision categories (保留/改写/合并/人工确认), specific config.yaml keys, and concrete rules (伪对立把 B 放入主干; 同义递进合并), with only minor gaps such as no worked before/after example inline.	4 / 5
Workflow Clarity	A clear 7-step sequence exists with explicit validation checkpoints (two self-eval rounds with distinct scopes, 原句—改写句—不变量对照) and a feedback loop (无法证明零损失则保留原句并标记人工确认), but the dual numbered-list structure makes the overall flow slightly less crisp than level 5.	4 / 5
Progressive Disclosure	The body is a concise overview that points via clearly signaled one-level-deep links to two real reference files (references/machine-patterns.md '模式与处置', references/regression-cases.md '匿名回归样例'), with detailed material appropriately split out rather than inlined.	5 / 5
	Total	17 / 20 Passed

Description

62%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 is domain-specific and lists concrete, comprehensive targets, but lacks an explicit positive 'Use when...' trigger and relies on somewhat generic action verbs. It is clearly distinct from other skills but weaker on trigger guidance and natural-phrase coverage.

Suggestions

Add an explicit 'Use when...' clause naming natural trigger phrases (e.g., 'Use when polishing NSFC grant proposals to remove AI tone, reduce machine flavor, or humanize 标书 wording').

Include colloquial synonyms users actually say (去AI味/降AI率/润色标书) alongside 'AI 机器味' to broaden natural trigger coverage.

Vary the action verbs beyond 去除/处理/覆盖 to concrete operations (改写、合并、释义、归位) to strengthen specificity.

Dimension	Reasoning	Score
Specificity	Lists specific targets across four layers (词语/句法/段落/章节) and concrete pattern types (伪对立、工程协议腔、规格书式字段串、术语漂移、边界声明过重、研究动作不清), but the action verbs (去除/处理/覆盖) are generic rather than varied concrete actions, so it stops short of level 5.	4 / 5
Completeness	The 'what' is clear (去除 NSFC 标书 AI 机器味, covering four layers and six pattern types) but there is no positive 'Use when...' trigger clause; only exclusions (不适用：非标书内容/需修改格式/需补充新内容) are given, which per the guideline caps completeness at 3.	3 / 5
Trigger Term Quality	Natural trigger terms like 'NSFC 标书' and 'AI 机器味' are present, but common variations a user might say ('去AI味', '降AI率', '润色标书') are missing and the remaining pattern names are technical rather than colloquial.	3 / 5
Distinctiveness Conflict Risk	The NSFC grant-proposal AI-flavor removal niche with a specific four-layer/six-pattern taxonomy is highly distinctive and very unlikely to trigger for an unrelated skill, fitting the clear-niche anchor.	5 / 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
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: huangwb8/ChineseResearchLaTeX
Path: skills/nsfc-humanization/SKILL.md
Commit: dd19421

Reviewed: 4 days 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.