neat-freak

Knowledge and governance closeout: reconcile project docs, rule files (CLAUDE.md/AGENTS.md), authorized agent memory, and workspace residue with what the code and runtime actually do, so the next session or the next person starts from one current answer. Trigger when the user names "neat-freak", "洁癖", or "/neat" — and also on clear knowledge-closeout intent without the name: syncing or tidying project docs/rules/memory after development ("把文档和记忆整理一下", "收尾时把文档同步掉", "docs 和代码对不上了"), stale or conflicting CLAUDE.md/memory, a clean handoff to a teammate or a fresh session, or auditing whether workspace rules are actually followed. Do not trigger for pure coding/refactoring/debugging tasks, tidying data or prose (JSON, 周报, changelog announcements), or a bare "整理" with no project-knowledge context.

Quality

92%

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

85%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.

A well-engineered instruction skill: two clearly sequenced paths with explicit status vocabularies, validation gates, a report template, and a genuinely progressive reference structure that checks out on disk. The recurring costs are some repetition of the cleanup-confirmation rule and a handful of principle-level steps that could carry one concrete example command.

Suggestions

State the destructive-cleanup confirmation rule once in a single authoritative place (e.g., the 权限和范围 section) and reference it from step 6, step 7, and the checklist, rather than restating it nearly verbatim four times.

Add one concrete example command or search pattern to principle-level steps, e.g. a sample rg invocation under "路由受影响知识面" for finding retired symbols/old field names.

Tighten the light-path step 3 rule-file guidance by pointing to the five required elements inline or in a reference instead of re-deriving them in the checklist as well.

Dimension	Reasoning	Score
Conciseness	The body is dense and assumes Claude's competence throughout — no space is spent explaining what git, CLAUDE.md, or memory systems are, and guidance is packed into tables and terse rules. However, the destructive-cleanup confirmation rule (report first, delete only after explicit user confirmation after the report) is stated nearly verbatim in "权限和范围先于洁癖", step 6, step 7, and twice in the final checklist, which is trimmable repetition.	4 / 5
Actionability	Mostly concrete and executable: a runnable command (`bash scripts/audit-inventory.sh <project-root>`), an explicit fact-status vocabulary (`verified-current`/`changed-and-verified`/`pending`/`out-of-scope`/`not-applicable`), a fixed diff-record format (`source of truth → stale surfaces → intended action → verification`), a copy-paste report template, and a final checklist. A few steps remain principle-level ("先减后加地修改", "根据改动类型搜索旧字段") without naming concrete search commands or examples, keeping it below fully-executable.	4 / 5
Workflow Clarity	Sequencing is explicit and validated: a 5-step light path, a 0–7 full path, validation gates in step 6 (run existing gates: doc links, lint, test, build; "不要为了过门禁注释掉错误"), a confirm-then-cleanup-then-re-audit feedback loop for the destructive cleanup, and a final self-check checklist. Validation for the destructive/batch operations is present, so the destructive-cap does not apply; this matches the top anchor with checkpoints and error-recovery loops.	5 / 5
Progressive Disclosure	The body is a well-structured overview that delegates detail to four one-level-deep, clearly-signaled references ([references/agent-paths.md], [references/governance.md], [references/sync-matrix.md], [references/verification.md]) — all present on disk, each introduced inline at the point of use, summarized in a "参考资料" section, with no further skill-file nesting inside the references. Detail is appropriately split and navigation is easy, matching the top anchor.	5 / 5
	Total	18 / 20 Passed

Description

95%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.

A strong description: it states what the skill does over a concrete set of artifacts and gives unusually thorough trigger guidance in both English and Chinese, including negative triggers that sharply reduce misfires. The only soft spot is that capability coverage is expressed as one reconcile action over several surfaces rather than a list of distinct concrete operations.

Suggestions

Consider enumerating the distinct operations performed on each surface (e.g., "rewrite stale docs in place, create minimal rule files, list deletion candidates, re-audit after cleanup") instead of the single verb "reconcile" to reach top-anchor specificity.

Dimension	Reasoning	Score
Specificity	The description names concrete targets and actions: "reconcile project docs, rule files (CLAUDE.md/AGENTS.md), authorized agent memory, and workspace residue with what the code and runtime actually do". It covers several specific surfaces but is essentially one broad reconcile action over multiple artifacts rather than the multiple distinct concrete actions (extract, fill, merge, convert) of the top anchor, so it sits noticeably above the midpoint without fully matching anchor 5.	4 / 5
Completeness	Both what ("Knowledge and governance closeout: reconcile project docs, rule files... and workspace residue with what the code and runtime actually do, so the next session or the next person starts from one current answer") and when ("Trigger when the user names... and also on clear knowledge-closeout intent without the name: syncing or tidying project docs/rules/memory after development...") are explicitly and concretely answered, matching the top anchor.	5 / 5
Trigger Term Quality	Trigger coverage is comprehensive including synonyms and both English and Chinese natural phrases: "neat-freak", "洁癖", "/neat", "把文档和记忆整理一下", "docs 和代码对不上了", "stale or conflicting CLAUDE.md/memory", "clean handoff", "auditing whether workspace rules are actually followed", plus explicit negative triggers ("Do not trigger for pure coding/refactoring/debugging tasks", "a bare 整理"). This matches the anchor for comprehensive natural-term coverage with synonyms.	5 / 5
Distinctiveness Conflict Risk	The niche (knowledge/governance closeout) is clearly distinct and the explicit do-not-trigger list ("pure coding/refactoring/debugging tasks, tidying data or prose (JSON, 周报, changelog announcements), or a bare 整理") actively removes the overlap risk with general coding or cleanup skills, matching the clear-niche, minimal-conflict anchor.	5 / 5
	Total	19 / 20 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: KKKKhazix/khazix-skills
Path: neat-freak/SKILL.md
Commit: 322346d

Reviewed: 3 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.