Anthropic의 엔지니어링 블로그 에서 소개된 "생각" 도구 개념을 기반으로 한 법적 추론 및 분석을 위한 엔터프라이즈급 MCP 서버입니다.
Cerebra Legal은 법적 추론 및 분석을 위한 세 가지 강력한 도구를 제공합니다.
legal_think - 도메인별 지침과 템플릿을 사용하여 복잡한 법률 문제를 분석하는 데 도움이 되는 체계적인 법적 추론 도구입니다.
legal_ask_followup_question - 도메인별 옵션을 제공하여 법률적 맥락에서 후속 질문을 묻는 데 특화된 도구입니다.
legal_attempt_completion - 적절한 구조와 인용 형식을 사용하여 법률 분석 결과를 제시하는 도구입니다.
서버는 합법적인 도메인(ANSC 이의 제기, 소비자 보호, 계약 분석)을 자동으로 감지하고 도메인별 지침, 템플릿 및 피드백을 제공합니다.
Related MCP server: Codebase Intelligence MCP Server
도메인 감지 : 분석의 합법적 도메인을 자동으로 식별합니다.
도메인별 지침 : 다양한 법률 도메인에 맞는 맞춤형 지침을 제공합니다.
구조화된 템플릿 : 법률 분석을 위한 도메인별 템플릿을 제공합니다.
인용 형식 : 법적 인용을 적절하게 형식화합니다.
사고 품질 분석 : 법적 추론 품질에 대한 피드백을 제공합니다.
수정 지원 : 이전 생각을 수정할 수 있습니다.
지엑스피1
npm start저장소에는 서버와 상호 작용하는 방법을 보여주는 테스트 클라이언트가 포함되어 있습니다.
# Make the test client executable
chmod +x test-client.js
# Run the test client
./test-client.js테스트 클라이언트는 다음을 수행합니다.
서버를 시작합니다
사용 가능한 도구를 얻으려면 도구/목록 요청을 보내세요
샘플 생각을 담은 legal_think 요청을 보내세요
서버의 응답을 표시합니다
Claude에 서버를 추가하려면 다음 구성으로 MCP 설정 파일을 업데이트하세요.
~/.config/Code/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.json 파일을 편집합니다.
{
"mcpServers": {
"cerebra-legal": {
"command": "node",
"args": ["/path/to/mcp-cerebra-legal-server/build/index.js"],
"disabled": false,
"alwaysAllow": []
}
}
}~/Library/Application Support/Claude/claude_desktop_config.json (macOS) 또는 플랫폼에 맞는 파일을 편집하세요.
{
"mcpServers": {
"cerebra-legal": {
"command": "node",
"args": ["/path/to/mcp-cerebra-legal-server/build/index.js"],
"disabled": false,
"alwaysAllow": []
}
}
}Claude에 서버를 추가하면 대화에서 도구를 사용할 수 있습니다.
이러한 법률 도구와 AI 간의 최적의 상호작용을 위해서는 시스템 프롬프트에 적절한 안내를 포함하는 것이 필수적입니다. 이러한 도구는 표준 Cline 도구와 동일한 AI → 인간 상호작용 모델을 따르지만, 분야별 개선 사항이 적용됩니다.
시스템 프롬프트에는 각 도구에 대한 자세한 설명서가 포함되어야 합니다.
## legal_think
Description: A structured legal reasoning tool that helps analyze complex legal issues with domain-specific guidance and templates. This tool enables step-by-step legal analysis with feedback on reasoning quality.
Parameters:
- thought: (required) The current legal reasoning step
- thoughtNumber: (required) Current thought number in sequence
- totalThoughts: (required) Estimated total thoughts needed
- nextThoughtNeeded: (required) Whether another thought step is needed
- category: (optional) Legal domain category (auto-detected if not provided)
- references: (optional) Array of legal references
- isRevision: (optional) Whether this revises previous thinking
- revisesThoughtNumber: (optional) Which thought is being reconsidered
- requestGuidance: (optional) Whether to request domain-specific guidance
- requestTemplate: (optional) Whether to request a domain-specific template
Usage:
<legal_think>
<thought>Your legal reasoning step here</thought>
<thoughtNumber>Current thought number</thoughtNumber>
<totalThoughts>Estimated total thoughts</totalThoughts>
<nextThoughtNeeded>true or false</nextThoughtNeeded>
<category>Legal domain (optional)</category>
<references>Array of references (optional)</references>
<isRevision>true or false (optional)</isRevision>
<revisesThoughtNumber>Thought number being revised (optional)</revisesThoughtNumber>
<requestGuidance>true or false (optional)</requestGuidance>
<requestTemplate>true or false (optional)</requestTemplate>
</legal_think>
## legal_ask_followup_question
Description: Ask the user a legal domain-specific question to gather additional information needed to complete the task. This tool enhances the standard ask_followup_question with legal domain detection, terminology formatting, and domain-specific suggested options.
Parameters:
- question: (required) The question to ask the user. This will be automatically enhanced with appropriate legal terminology.
- options: (optional) An array of 2-5 options for the user to choose from. If not provided, domain-specific options will be automatically suggested.
- context: (optional) Additional context to help with domain detection and question formatting.
Usage:
<legal_ask_followup_question>
<question>Your question here</question>
<options>
Array of options here (optional), e.g. ["Option 1", "Option 2", "Option 3"]
</options>
<context>Additional context to help with domain detection (optional)</context>
</legal_ask_followup_question>
## legal_attempt_completion
Description: Present the result of your work to the user with proper legal structure and formatting. This tool enhances the standard attempt_completion with legal domain detection, document structuring, and citation formatting.
Parameters:
- result: (required) The result of the task. This will be automatically formatted with proper legal structure.
- command: (optional) A CLI command to execute to show a live demo of the result to the user.
- context: (optional) Additional context to help with domain detection and result formatting.
Usage:
<legal_attempt_completion>
<result>
Your final result description here
</result>
<command>Command to demonstrate result (optional)</command>
<context>Additional context to help with domain detection (optional)</context>
</legal_attempt_completion>이 지침은 AI가 다음 사항을 이해하도록 보장합니다.
이는 표준 도구의 특수 버전입니다.
동일한 AI → 인간 상호작용 흐름을 유지합니다.
추가 기능과 매개변수가 있습니다.
도구 호출을 올바르게 형식화하는 방법
이러한 지침이 없으면 AI가 해당 도구에 내장된 도메인별 기능을 최대한 활용하지 못할 수도 있습니다.
legal_think 도구는 체계적인 사고를 통해 복잡한 법률 문제를 분석하는 데 도움이 됩니다.
I need to analyze an ANSC contestation where a claimant argues that technical specifications in a tender were too restrictive.Claude는 legal_think 도구를 사용하여 다음을 수행합니다.
법적 도메인 감지(ANSC 이의 제기)
도메인별 지침 제공
분석을 위한 구조화된 템플릿 제공
법적 추론의 질에 대한 피드백을 제공하세요
이전 생각의 수정을 지원합니다
클로드가 법적 분석을 완료하기 위해 추가 정보가 필요한 경우:
What specific provisions of the technical specifications are being challenged?Claude는 legal_ask_followup_question 도구를 사용하여 다음을 수행합니다.
적절한 법률 용어로 질문을 구성하세요.
사용자가 선택할 수 있는 도메인별 옵션 제공
컨텍스트 인식 질문을 위한 법적 도메인 감지
클로드가 최종 법적 분석을 제시할 준비가 되었을 때:
Based on my analysis, the technical specifications requiring "minimum 5 years experience" appear disproportionate and likely violate Article 33(2) of Law 131/2015 on public procurement.Claude는 legal_attempt_completion 도구를 사용하여 다음을 수행합니다.
적절한 법적 구조로 결론을 구성하세요
법률 인용문 추출 및 형식 지정
분석을 명확한 섹션으로 구성하세요
전문적인 법률 문서 형식을 제공합니다
{
"thought": "Analyzing ANSC contestation where claimant argues technical specifications were too restrictive.",
"thoughtNumber": 1,
"totalThoughts": 5,
"nextThoughtNeeded": true,
"category": "ansc_contestation", // Optional, auto-detected if not provided
"references": ["Law 131/2015", "ANSC Decision #12345"], // Optional
"isRevision": false, // Optional
"revisesThoughtNumber": null, // Optional
"requestGuidance": true, // Optional
"requestTemplate": true // Optional
}{
"question": "What specific provisions of the technical specifications are being challenged?",
"options": [ // Optional, auto-generated if not provided
"Are you challenging the experience requirements?",
"Are you challenging the technical capacity requirements?",
"Are you challenging the financial requirements?",
"Are you challenging the certification requirements?"
],
"context": "ANSC contestation regarding procurement of IT equipment" // Optional
}{
"result": "Based on the analysis of ANSC contestation #12345, the technical specifications requiring 'minimum 5 years experience' appear disproportionate and likely violate Article 33(2) of Law 131/2015 on public procurement.",
"command": null, // Optional
"context": "ANSC contestation analysis" // Optional
}서버는 모듈식 아키텍처로 구축되었습니다.
도메인 감지기 : 분석의 법적 도메인을 식별합니다.
법률 지식 기반 : 도메인별 지침 및 템플릿 제공
인용 형식 지정기 : 법적 인용을 적절하게 형식화합니다.
도구 구현 : 각 도구에 대한 논리를 처리합니다.
mcp-cerebra-legal-server/
├── src/
│ ├── shared/ # Shared components
│ │ ├── DomainDetector.ts
│ │ ├── LegalKnowledgeBase.ts
│ │ ├── CitationFormatter.ts
│ │ └── types.ts
│ ├── tools/ # Tool implementations
│ │ ├── LegalThinkTool.ts
│ │ ├── LegalAskFollowupQuestionTool.ts
│ │ └── LegalAttemptCompletionTool.ts
│ ├── utils/ # Utilities
│ │ └── logger.ts
│ └── index.ts # Main server entry point
├── build/ # Compiled JavaScript
├── test-client.js # Test client
├── package.json
└── tsconfig.jsonnpm run build# Run the test client
./test-client.js이 프로젝트는 GitHub에서 https://github.com/yoda-digital/mcp-cerebra-legal-server 에서 사용할 수 있습니다.
"생각" 도구: 클로드가 복잡한 도구 사용 상황에서 멈추고 생각할 수 있도록 지원 - Anthropic Engineering Blog
MIT
A specialized tool for asking follow-up questions in legal contexts. This tool helps gather additional information needed for legal analysis by formulating precise questions with domain-specific options.
When to use this tool:
When you need additional information to complete a legal analysis
When clarification is needed on specific legal points
When gathering evidence or documentation for a legal case
When exploring alternative legal interpretations
Key features:
Automatic detection of legal domains
Domain-specific question suggestions
Legal terminology formatting
Structured options for efficient information gathering
| Name | Required | Description | Default |
|---|---|---|---|
| question | Yes | The legal question to ask the user | |
| options | No | An array of 2-5 options for the user to choose from (optional) | |
| context | No | Additional context about the legal issue (optional) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It lists 'Key features' like 'Automatic detection of legal domains' and 'Legal terminology formatting,' which hint at functionality, but fails to disclose critical behavioral traits such as whether this tool modifies data, requires specific permissions, has rate limits, or what the output looks like. For a tool with no annotations, this is a significant gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-structured with clear sections ('When to use this tool,' 'Key features'), making it easy to scan. It's appropriately sized with no redundant sentences, though it could be slightly more concise by integrating some points. Every sentence adds value, such as explaining domain-specific options.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity (legal domain tool with 3 parameters, no annotations, no output schema), the description is partially complete. It covers purpose and usage well but lacks details on behavioral aspects and output. Without annotations or an output schema, the description should do more to explain what happens when the tool is invoked, but it provides enough context for basic use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents all three parameters (question, options, context) with descriptions. The description adds no additional meaning beyond the schema, such as examples or usage tips for parameters. The baseline score of 3 is appropriate since the schema does the heavy lifting, but the description doesn't compensate or enhance understanding.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose as 'asking follow-up questions in legal contexts' and 'gather additional information needed for legal analysis,' which is specific and actionable. However, it doesn't explicitly differentiate from sibling tools like legal_attempt_completion or legal_think, leaving some ambiguity about when to choose this tool over alternatives.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description includes a 'When to use this tool' section with four clear scenarios (e.g., 'When you need additional information to complete a legal analysis'), providing explicit guidance on appropriate contexts. It lacks explicit exclusions or comparisons to sibling tools, but the scenarios are well-defined and helpful.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
A specialized tool for presenting legal analysis results and conclusions. This tool formats legal conclusions with proper structure, extracts and formats citations, and provides a professional legal document format.
When to use this tool:
When presenting the final results of a legal analysis
When summarizing legal findings and recommendations
When providing a structured legal opinion
When concluding a legal reasoning process
Key features:
Automatic detection of legal domains
Proper legal document formatting
Citation extraction and formatting
Structured sections for clear communication
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes | The legal analysis result or conclusion | |
| command | No | A CLI command to execute (optional) | |
| context | No | Additional context about the legal issue (optional) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden but provides minimal behavioral disclosure. It mentions formatting features but doesn't address permissions needed, whether it modifies data, rate limits, error conditions, or what the output looks like. For a tool with 3 parameters and no output schema, this leaves significant gaps in understanding how the tool actually behaves.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-structured with clear sections (purpose, usage guidelines, key features) and efficiently communicates core information. While slightly verbose in listing multiple 'when to use' examples, each sentence adds value and the overall length is appropriate for the tool's complexity.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given 3 parameters, no annotations, and no output schema, the description provides adequate purpose and usage context but lacks sufficient behavioral details. It explains what the tool does and when to use it, but doesn't adequately address how it works, what permissions are needed, or what the output format will be. This leaves gaps for a tool that presumably generates formatted legal documents.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents all parameters. The description adds no specific information about parameter usage, relationships, or examples beyond what's in the schema. The baseline of 3 is appropriate when the schema does the heavy lifting, though the description could have explained how parameters interact with the formatting features mentioned.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: 'presenting legal analysis results and conclusions' with specific functions like formatting, citation extraction, and document structuring. It distinguishes from sibling tools (legal_ask_followup_question, legal_think) by focusing on final presentation rather than analysis or questioning, though it doesn't explicitly name these alternatives.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The 'When to use this tool' section provides clear context: presenting final results, summarizing findings, providing structured opinions, and concluding legal reasoning. It implicitly distinguishes from siblings by focusing on completion rather than intermediate steps, but doesn't explicitly state when NOT to use it or name specific alternatives.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
A powerful tool for structured legal reasoning that helps analyze complex legal issues. This tool provides domain-specific guidance and templates for different legal areas including ANSC contestations, consumer protection, and contract analysis.
When to use this tool:
Breaking down complex legal problems into structured steps
Analyzing legal requirements and compliance
Verifying that all elements of a legal test are addressed
Building comprehensive legal arguments with proper citations
Key features:
Automatic detection of legal domains
Domain-specific guidance and templates
Support for legal citations and references
Revision capabilities for refining legal arguments
Thought quality feedback
| Name | Required | Description | Default |
|---|---|---|---|
| thought | Yes | The main legal reasoning content | |
| category | No | Category of legal reasoning (optional, will be auto-detected if not provided) | |
| references | No | References to laws, regulations, precedents, or previous thoughts (optional) | |
| isRevision | No | Whether this thought revises a previous legal reasoning (optional) | |
| revisesThoughtNumber | No | The thought number being revised (if isRevision is true) | |
| requestGuidance | No | Set to true to receive domain-specific legal guidance | |
| requestTemplate | No | Set to true to receive a template for this type of legal reasoning | |
| thoughtNumber | Yes | Current thought number | |
| totalThoughts | Yes | Estimated total thoughts needed | |
| nextThoughtNeeded | Yes | Whether another thought step is needed |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. It mentions key features like automatic domain detection, guidance/templates, citation support, revision capabilities, and thought quality feedback, which gives useful context about how the tool behaves. However, it doesn't address important behavioral aspects like whether this is a read-only analysis tool or if it modifies data, what permissions might be needed, or any rate limits.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-structured with clear sections (purpose, when to use, key features) and each sentence adds value. It's appropriately sized for a complex tool with 10 parameters, though the 'Key features' section could be more concise as some items overlap with earlier content.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a complex legal reasoning tool with 10 parameters and no annotations or output schema, the description provides good purpose and usage context but has significant gaps. It doesn't explain what the tool outputs (no output schema), doesn't address behavioral constraints, and while it mentions legal domains, it doesn't fully explain how the complex parameter set works together for the reasoning process.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents all 10 parameters thoroughly. The description doesn't add any specific parameter information beyond what's in the schema - it mentions general capabilities like domain detection and revision, but doesn't explain how parameters like 'category' or 'isRevision' relate to these features. Baseline 3 is appropriate when the schema does the heavy lifting.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose as 'structured legal reasoning that helps analyze complex legal issues' and mentions specific legal domains like ANSC contestations, consumer protection, and contract analysis. It distinguishes itself from siblings by focusing on reasoning rather than follow-up questions or completion attempts, though it doesn't explicitly contrast with them.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The 'When to use this tool' section provides clear guidance on four specific scenarios like breaking down complex legal problems and analyzing legal requirements. It doesn't mention when NOT to use it or explicitly name sibling tools as alternatives, but the context is well-defined for legal reasoning tasks.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
legal_ask_followup_questionlegal_attempt_completionlegal_thinkScored across 3 tools
Each tool has a clearly distinct purpose: legal_think is for structured reasoning and analysis, legal_ask_followup_question is for gathering additional information, and legal_attempt_completion is for presenting final conclusions. The descriptions reinforce these distinct roles with no overlap in core functionality.
All tools follow a consistent 'legal_' prefix with descriptive action names (think, ask_followup_question, attempt_completion), using snake_case uniformly. This pattern makes the tool set predictable and easy to navigate.
With only 3 tools, the set feels thin for a legal analysis domain, potentially lacking operations like document retrieval, case law lookup, or specific legal research functions. While the tools cover reasoning, questioning, and completion, the scope suggests more tools might be needed for comprehensive legal workflows.
The tool set is severely incomplete for legal analysis, missing essential operations such as accessing legal databases, retrieving statutes or case law, validating citations, or drafting legal documents. The existing tools focus only on internal reasoning and presentation, leaving significant gaps that would hinder an agent's ability to perform thorough legal tasks.
Hybrid human + AI expertise for faster, trusted answers and decisions via MCP Server.
Connect AI to millions of laws and court cases with the Lawstronaut MCP.
AI governance MCP server for EU AI Act compliance and jurisdiction verification
