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clear审计未展示

legal-review法律审查

Agent Skill

legal-review 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

441

周安装

18

GitHub Stars

公开资料未说明

下载量

143
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

MIT

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:legal-review(法律审查)
来源仓库:https://github.com/front-depiction/claude-setup
仓库路径:skills/legal-review
安装命令:
npx skills add front-depiction/claude-setup --skill "legal-review"
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。

AgentSkills.tonpx skills
npx skills add front-depiction/claude-setup --skill "legal-review"

简介

辅助法律文档审查与条款分析,提升审核效率。

  • 适用于 Codex、Claude 等平台中的法务协作场景。
  • 通过 GitHub 安装,使用 npx skills add front-depiction/claude-setup --skill "legal-review"。
  • 注意保密要求,禁止处理敏感客户信息。
  • 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

name
legal-review
description
Orchestrate lawyer agents to review code for compliance with codebase laws. Spawns counsel in parallel to produce a unified legal brief.

Legal Review

Orchestrates legal counsel to review code for compliance with codebase laws and standards. Determines applicable laws based on file patterns, spawns lawyers in PARALLEL, then aggregates findings into a unified legal brief.

Execution Protocol

You are the Legal Review Coordinator. Follow this protocol exactly:

Phase 1: Resolve Scope

Interpret natural language arguments to determine what files to review.

interpretScope :: $ARGUMENTS -> Scope
interpretScope args
  | null args                                    = Staged
  | isFilePath args                              = Files [args]
  | mentions "staged"                            = Staged
  | mentions "unstaged" || mentions "diff"       = Diff
  | mentions "all" || mentions "entire" ||
    mentions "everything" || mentions "codebase" = All
  | mentions "today" || mentions "changed"       = ChangedToday
  | otherwise                                    = Search (extractIntent args)

Natural Language Patterns:

PatternInterpretationResolution Strategy
the chat componentComponent searchGlob for **/chat*.ts, **/Chat*.tsx
all VM files in the UI appScoped patternGlob for apps/ui/**/*.vm.ts
my staged changesGit stagedgit diff --cached --name-only
the sidebar and navigationMultiple componentsGlob for each term
everything I changed todayRecent changes`git diff --name-only $(git log --since="midnight" --format=%Htail -1)^`
the user authentication flowDomain searchGrep for auth patterns, Glob for auth files

Execute scope resolution based on interpretation:

# For Staged (default, or "my staged changes"):
git diff --cached --name-only

# For Diff ("unstaged changes", "my diff"):
git diff --name-only

# For ChangedToday ("everything I changed today", "today's changes"):
git log --since="midnight" --name-only --format="" | sort -u

# For Files (explicit path like "src/components/Chat.vm.ts"):
# Use the provided file path directly

# For All ("the entire codebase", "all files"):
# Use Glob to find all *.ts and *.tsx files

# For Search (component/domain references):
# Use Glob and Grep to find relevant files

Search Resolution Examples:

When the user says "the chat component":

# Try multiple patterns to find relevant files
glob "**/chat*.ts" "**/Chat*.tsx" "**/chat/**"
grep -l "Chat" --include="*.ts" --include="*.tsx"

When the user says "all VM files in the UI app":

glob "apps/ui/**/*.vm.ts"

When the user says "the sidebar and navigation components":

glob "**/sidebar*.ts" "**/Sidebar*.tsx" "**/navigation*.ts" "**/Navigation*.tsx"
glob "**/nav*.ts" "**/Nav*.tsx"

When the user says "the user authentication flow":

grep -l "auth" "login" "session" --include="*.ts" --include="*.tsx"
glob "**/auth/**" "**/login/**" "**/session/**"

Phase 2: Determine Applicable Laws

Match files to applicable bodies of law based on patterns:

Body of LawPatternSkillStatus
VM Architecture Laws*.vm.tsthe-vm-standardACTIVE
Effect Pattern Laws*.ts (Effect code)effect-patterns-lawFUTURE
Schema Laws**/Schema*.ts, **/domain/**schema-lawFUTURE
Layer Composition Laws**/Layer*.ts, **/*Layer.tslayer-lawFUTURE
Error Handling Laws**/errors/**, **/*Error.tserror-handling-lawFUTURE
Testing Laws*.test.ts, *.spec.tstesting-lawFUTURE
determineApplicableLaws :: [FilePath] -> [(Law, [FilePath])]
determineApplicableLaws files = do
  let vmFiles = filter (matches "*.vm.ts") files
  -- Add more laws here as they become active
  catMaybes
    [ guard (not (null vmFiles)) >> pure (VMArchitectureLaw, vmFiles)
    ]

If no files match any active law jurisdiction, report: "No files within active law jurisdiction. Review complete."

Phase 3: Dispatch Lawyer in PARALLEL

For each body of law with applicable files, spawn a lawyer agent IN PARALLEL.

CRITICAL: Spawn all applicable lawyers simultaneously using parallel agent dispatch.

Lawyer Agent

Spawn the /lawyer agent with appropriate context:

<lawyer-agent-prompt> You are legal counsel for this codebase. Your task is to review the following files for compliance with applicable laws.

Files under review: $FILES

Applicable laws based on file patterns: $APPLICABLE_LAWS

Your authority: You are the /lawyer agent. Load relevant law skills for the full specifications.

Procedure:

  1. Determine which laws apply to each file based on patterns
  2. Load each applicable law skill (e.g., /the-vm-standard for *.vm.ts files)
  3. Read each file completely
  4. Audit against all applicable statutes systematically
  5. Issue formal legal opinion with charges for any violations

Required output format: For each file, produce a verdict with:

  • File path
  • Applicable laws
  • Compliance status (COMPLIANT | VIOLATION | ADVISORY)
  • For violations: charges citing specific law, statute, line number, evidence, and remedy

</lawyer-agent-prompt>

Phase 4: Await and Aggregate

After all lawyers complete, aggregate findings:

data Finding = Finding
  { law       :: Law              -- e.g., VMArchitectureLaw
  , file      :: FilePath
  , severity  :: Severity         -- Critical | Major | Minor | Advisory
  , statute   :: String           -- e.g., "VM-COVENANT-VII" or "EFFECT-PATTERN-III"
  , violation :: Description
  , evidence  :: (LineNumber, CodeSnippet)
  , remedy    :: Recommendation
  }

data Verdict
  = Approved            -- no Critical or Major findings
  | ConditionalApproval -- Major findings exist
  | Blocked             -- Critical findings must be resolved

verdict :: [Finding] -> Verdict
verdict findings
  | any ((== Critical) . severity) findings = Blocked
  | any ((== Major) . severity) findings    = ConditionalApproval
  | otherwise                                = Approved

Phase 5: Produce Legal Brief

Output the unified legal brief:

===============================================================================
                          LEGAL BRIEF
===============================================================================

SCOPE: [Staged | Diff | Files [...] | All]
FILES REVIEWED: [count] files
LAWS APPLIED: [list of bodies of law consulted]

-------------------------------------------------------------------------------
                        EXECUTIVE SUMMARY
-------------------------------------------------------------------------------

| Severity | Count |
|----------|-------|
| Critical | N     |
| Major    | N     |
| Minor    | N     |
| Advisory | N     |

VERDICT: [Approved | Conditional Approval | Blocked]

-------------------------------------------------------------------------------
                        COUNSEL OPINIONS
-------------------------------------------------------------------------------

## VM Architecture Laws

[Full opinion for VM files]

## [Other Laws - when active]

[Full opinion for other jurisdictions]

-------------------------------------------------------------------------------
                        RECOMMENDATIONS
-------------------------------------------------------------------------------

[Prioritized remediation steps, Critical first, then Major]

===============================================================================
                        BRIEF CONCLUDED
===============================================================================

Usage Examples

# Review staged changes (default)
/legal-review
/legal-review my staged changes

# Review specific file (still supported)
/legal-review src/components/Chat.vm.ts

# Review by component name
/legal-review the chat component
/legal-review the sidebar and navigation components

# Review by domain/feature
/legal-review the user authentication flow
/legal-review everything related to sessions

# Review scoped to an app or package
/legal-review all VM files in the UI app
/legal-review the services in packages/ai

# Review by time
/legal-review everything I changed today
/legal-review my recent changes

# Review unstaged changes
/legal-review my unstaged changes
/legal-review the current diff

# Review entire codebase (expensive)
/legal-review the entire codebase
/legal-review all files

Extending: Adding New Bodies of Law

To add a new body of law:

  1. Define the law - Create .claude/skills/{law-name}/SKILL.md with the statutes
  2. Add to jurisdiction table - Add pattern and skill name to Phase 2 table above
  3. Activate - Change status from FUTURE to ACTIVE in jurisdiction table

The lawyer agent (/lawyer) automatically loads applicable law skills based on file patterns. The lawyer's jurisdiction table must also be updated to match.

Severity Definitions

SeverityDefinitionMerge Impact
CriticalFundamental violation that will cause runtime issues or architectural decayBLOCKS merge
MajorSignificant deviation from standards that should be fixedShould fix before merge
MinorStyle or convention issuesFix when convenient
AdvisorySuggestions for improvementOptional

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

04

需要参考平台分布和安装热度时

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

补充不同宿主或平台的使用分布数据

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

Claude Code

30.86%
按下载量换算44

windsurf

23.55%
按下载量换算34

trae

17.11%
按下载量换算24

OpenCode

12.42%
按下载量换算18

Codex

8.53%
按下载量换算12

Antigravity

3.68%
按下载量换算5

安全审计

暂无安全审计结果可展示。

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 front-depiction/claude-setup --skill "legal-review" 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。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