Token导航 LogoToken导航TokenDH.com
研究检索执行命令github未标认证来源可访问许可证需确认审计提醒

ln-162-skill-reviewerln 162 技能审核员

Agent Skill

ln-162-skill-reviewer 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

2,496

周安装

104

GitHub Stars

441

下载量

832
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:ln-162-skill-reviewer(ln 162 技能审核员)
来源仓库:https://github.com/levnikolaevich/claude-code-skills
仓库路径:skills/ln-162-skill-reviewer
安装命令:
npx skills add https://github.com/levnikolaevich/claude-code-skills --skill ln-162-skill-reviewer
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/levnikolaevich/claude-code-skills --skill ln-162-skill-reviewer

简介

ln-162-skill-reviewer 用于查找、检索和筛选相关信息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词或任务场景快速定位候选结果。
  • 通过 npx skills add 命令从指定 GitHub 仓库安装,需结合原始 README 核验具体用法。
  • 安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写操作。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Paths: File paths (shared/, references/, ../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root. If shared/ is missing, fetch files via WebFetch from https://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/{path}.

ln-162-skill-reviewer

Type: L3 Worker (standalone-capable) Category: 1XX Documentation Pipeline

Universal reviewer with two modes:

  • SKILL for ln-*/SKILL.md
  • COMMAND for .claude/commands/*.md
  • repo suite for this skills repository when repo / repo-checks is requested or the current repo is claude-code-skills
Plan Mode behavior: Phases 1-4 and 6-7 are research. Run them fully in Plan Mode, write the report plus fix list into the plan, and apply edits only after approval.

Mode Detection

ConditionModeReview Profile
ln-*/SKILL.md files exist in CWDSKILLFull D1-D11 + M1-M6
.claude/commands/*.md files existCOMMANDStructural + actionability
Both existSKILLOverride with $ARGUMENTS=commands

Input

$ARGUMENTS options:

  • empty -> auto-detect mode and scope
  • ln-400 ln-500 -> SKILL mode, specific skills
  • repo or repo-checks -> run canonical repo-specific checks after universal review
  • commands -> COMMAND mode, all .claude/commands/*.md
  • deploy.md run-tests.md -> COMMAND mode, specific files

When invoked by another skill, file paths may be passed directly.


SKILL Mode

Phase 1: Scope Detection

If $ARGUMENTS is provided:

  • treat each token as a skill directory prefix
  • glob {prefix}*/SKILL.md

If $ARGUMENTS is empty:

  • inspect git diff, staged files, and untracked files
  • collect primary skill dirs
  • collect affected skills referencing changed shared files
  • collect dependency skills from worker tables and Skill() invocations

Report: Scope: {N} primary, {M} affected, {K} dependency skills.

Phase 2: Automated Verification

Run:

# Never pass a giant mixed scope in one invocation.
# Split by `ln-NXX` family or by coordinator/worker group.
bash references/run_checks.sh {batch_1 SKILL.md files}
bash references/run_checks.sh {batch_2 SKILL.md files}

Batching rules:

  • keep each invocation within one ln-NXX family when possible
  • if scope mixes coordinators and workers, split them into separate invocations
  • if one family is still large, chunk it into 5-10 files per invocation

Automated failures are pre-verified. Record every one.

Phase 3: Structural Review

MANDATORY READ: Load references/structural_review.md, shared/references/skill_contract.md, shared/references/procedural_skill_sop_guide.md, shared/references/mcp_applicability_matrix.md, and shared/references/mcp_tool_preferences.md

Review every skill in scope across D1-D11.

Treat these as structural issues, not style nits:

  • missing **Type:** when role-sensitive checks depend on it
  • worker independence violations in L3 workers
  • broken shared paths
  • stale root-doc assumptions — CLAUDE.md duplicating AGENTS.md content instead of using the @AGENTS.md import stub with a bounded harness delta
  • markdown-analysis skills missing markdown_read_protocol.md
  • extraction or audit skills contradicting the shared docs-quality contract
  • skills contradicting the shared skill contract
  • skills that should make hex-line primary but only describe built-in file tools
  • skills that should make hex-graph primary but still describe grep/manual analysis as the default path
  • skills that add hex-graph or hex-line with no real applicability per the shared matrix
  • wrong MCP namespaces such as mcp__hex_graph__...
  • procedural skills with vague modal words, compound high-risk steps, missing point-of-use checklists, missing why for risky steps, missing executable preflight, or missing evidence

Phase 4: Intent Review

MANDATORY READ: Load references/intent_review.md

Apply M1-M6 to primary skills only. Read the git diff for each primary skill.

Phase 5: Fix

Auto-fix deterministic issues:

  • wrong paths
  • stale references
  • duplicated wording
  • worker-independence violations with exact removals
  • copied docs shell sections in command files

Do not guess on ambiguous behavior.

After fixes, re-read each primary skill end-to-end and compress redundant wording without changing behavior.

Phase 6: Report

Verdict rules:

  • any structural violation not auto-fixed -> FAIL
  • only advisory intent concerns remain -> PASS with CONCERNS
  • zero findings -> PASS

Report format:

## Skill Coherence Review -- {PASS|PASS with CONCERNS|FAIL}

**Scope:** {reviewed skills}
**Verdict:** {verdict}

Phase 7: Volatile Numbers Cleanup

Remove stale aggregate counts from SKILL.md files. Keep only local counts intrinsic to the reviewed file.


Repo Suite Mode

Use when running inside this repository or when $ARGUMENTS includes repo / repo-checks.

Phase 1: Universal Review

Run the applicable SKILL or COMMAND mode first. Do not skip the universal review.

Phase 2: Repo-Specific Checks

Run the canonical repo suite:

node skills-catalog/ln-162-skill-reviewer/references/repo_review_suite.mjs

The suite owns R1-R27. Do not copy those checks into host-specific command files.

Phase 3: Combined Report

Merge the universal review verdict and repo suite verdict:

| Source | Verdict | Details |
|--------|---------|---------|
| Universal review | {PASS|PASS with CONCERNS|FAIL} | {summary} |
| Repo suite | {PASS|PASS with WARNINGS|FAIL} | {R1-R27 summary} |
| Combined | {worst verdict} | |

List all FAIL and WARN items with file paths and fix descriptions.


COMMAND Mode

MANDATORY READ: Load references/command_review_criteria.md

Phase 1: Scope Detection

  • explicit file paths -> review those files
  • commands -> glob .claude/commands/*.md
  • coordinator-supplied file list -> review those files

Phase 2: Review

For each command file:

  • apply all command review criteria
  • verify source provenance
  • verify no copied docs shell sections remain

Phase 3: Fix

Auto-fix where safe:

  • missing frontmatter
  • missing allowed-tools
  • description too long
  • missing Last Updated
  • exact copied docs shell sections

Phase 4: Report

## Command Review -- {N} files

| File | Verdict | Issues |
|------|---------|--------|

Verdicts: PASS / FIXED / WARN / FAIL
Pass rate: {X}%

Rules

  • Automated checks are non-negotiable.
  • Read all scoped files before reporting.
  • Fix deterministic issues immediately.
  • Do not update versions or dates unless the user explicitly requests it.
  • shared/ changes affect all referencing skills.
  • Worker independence is mandatory for L3 workers:

- no **Coordinator:** - no **Parent:** - no required caller declaration

  • Docs-model drift is a structural defect, not a preference.
  • Unsupported platform APIs are prohibited outside clearly marked historical references.

Reference Files

  • references/structural_review.md
  • references/intent_review.md
  • references/automated_checks.md
  • references/run_checks.sh
  • references/run_runtime_suite.mjs
  • references/repo_review_suite.mjs
  • references/command_review_criteria.md
  • references/check_marketplace.mjs

Definition of Done

  • Scope detected
  • Automated checks executed
  • D1-D11 reviewed across all scoped skills
  • M1-M6 evaluated for primary skills
  • Fixable findings auto-fixed
  • Post-fix holistic compaction completed
  • Final verdict report generated

Version: 1.0.0 Last Updated: 2026-03-26

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.25%
按下载量换算302

Claude

31.09%
按下载量换算259

Cursor

18%
按下载量换算150

Gemini CLI

9.75%
按下载量换算81

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/levnikolaevich/claude-code-skills --skill ln-162-skill-reviewer 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills