p

persona-mcp

@SeoNaRu/persona-mcp
0 Stars 172 次浏览 SeoNaRu 更新于 2026-08-23
交流协作工具学术研究

MCP 服务配置

复制以下 JSON 到 OPClaw 或其他 MCP 客户端的配置文件中即可使用

{
  "mcpServers": {
    "persona-server": {
      "args": [
        "-m",
        "src.main"
      ],
      "command": "python3",
      "cwd": "/Users/seodong-ug/Desktop/persona_mcp"
    }
  }
}

服务介绍

Persona MCP Server

AI FGI (Focus Group Interview) MCP (Model Context Protocol) .

MCP ?

MCP (Model Context Protocol) AI .

MCP

  1. (Tools): AI
  2. (Resources): AI
  3. (Prompts):

MCP

Persona MCP Server FastMCP , :

  • ** (stdio) **: MCP JSON-RPC
  • ** **: asyncio
  • ** **: Pydantic /

?

MCP ** FGI ** . :

1.

  • AI
  • MBTI , , ,
  • (, , , )

2. FGI

  • (, , , , , )
  • AI

3.

  • " "
  •  ( -3)
  • SS, S, A, B, C

4.

  • /

5.

  • LLM ( )


  MCP Client       (Cursor AI, Claude Desktop )
  (Cursor AI)    

          JSON-RPC (stdio)
         

         Persona MCP Server                   
     
    FastMCP Framework                      
    - Tool Registration                    
    - Request/Response Handling            
     
                                             
     
    Tool Implementations                   
    - FGI Tools (fgi_tools.py)            
    - Predict (predict.py)                
    - Insights (insights.py)              
    - Freshness (freshness.py)            
    - Query (query.py)                    
    - Sample Data (sample_data.py)        
     
                                             
     
    Database Layer (SQLAlchemy)            
    - Persona Model                        
    - FGISurvey Model                      
    - FGIResponse Model                    
     

              

   SQLite Database         
   (data/fgi.db)           

              

   Groq LLM API            
   (AI   )

  1. FastMCP: MCP
  2. SQLAlchemy: ORM
  3. Groq API: LLM
  4. Pydantic:
  5. ** **:

1.

** : !**

   .        .


,        .

#     
questions = get_basic_profile_questions_tool()

** :**

  • age_range: () - 10, 20 , 20 , 30 , 30 , 40
  • gender: () - , , ,
  • location: () - , , , , , , ,
  • occupation_category: () - IT/, , , , , ,
  • income_range: () - 200 , 200-300, 300-500, 500 ,

** :** age_range, gender, location, occupation_category (4)
** :** income_range (1)

** :**

  1. get_basic_profile_questions_tool()
  3. update_basic_profile_tool()

 .

#     
check_basic_profile_tool()

#     
check_basic_profile_tool(persona_id="persona_123")

** :**

  • is_complete: (bool)
  • missing_fields:
  • message:
  • persona_id: ID ( None)

.

#    
update_basic_profile_tool(
    persona_id="persona_123",
    profile_data={
        "age_range": "20 ",        #     
        "gender": "",                 #     
        "location": "",               #     
        "occupation_category": "IT/", #     
        "income_range": "300-500"     # ,     
    }
)

:

  • profile_data get_basic_profile_questions_tool() .
  • ** 4(age_range, gender, location, occupation_category) .**
  • income_range .
  •   ( 25).
  • ** , .**

(AI )

#      
predict_persona_tool(
    responses=[
        {"question_id": "q1", "answer": "   ..."},
        {"question_id": "q2", "answer": "  ..."}
    ],
    user_id="user_123"
)

** :**

  • MBTI
  • MBTI

2. FGI

#   
create_fgi_survey_tool(
    category="technology",  # technology, lifestyle, culture, fashion, food, general
    persona_id="persona_123"  # 
)

** :**

  • technology:
  • lifestyle:
  • culture: /
  • fashion:
  • food:
  • general:

#      
generate_next_question_tool(
    survey_id="survey_123",
    previous_qa=[
        {"question": "     ?", "answer": "AI   "},
        {"question": "  ?", "answer": "   "}
    ],
    category="technology",
    max_questions=10  # :    
)

:

  • AI
  • (info_completeness)
  • (is_final)

#      
submit_fgi_response_tool(
    persona_id="persona_123",
    survey_id="survey_123",
    responses=[
        {"question_id": "q1", "answer": " "},
        {"question_id": "q2", "answer": " "}
    ]
)

3.

" " .

** :**

  • : 25
  • : 10-50 ( )
  • : 50-100

** :**

  • :  -3
  • 0
SS 500 -
S 300-499 -
A 200-299 -
B 100-199 -
C 50-99 -
F 0-49 - 
#  
check_freshness_tool(persona_id="persona_123")

** :**

  • (SS, S, A, B, C, F)

4.

#  /   
analyze_preference_tool(
    query=" ",
    category="fashion",  # 
    age_range="20 ",  # 
    gender="",  # 
    location="",  # 
    limit=20
)

** :**

  • /

#    
get_trend_insights_tool(
    topic="",
    persona_filters={
        "age_range": "20",
        "gender": ""
    },
    min_freshness=100  #   
)


#    
search_responses_tool(
    keyword="",
    category="fashion",
    limit=20
)

5.

get_persona_tool(persona_id="persona_123")

** :**

  • (, , )
  • MBTI
  • ( , )

list_personas_tool(
    age_range="20 ",
    gender="",
    location="",
    occupation_category="IT/",
    mbti_type="INTJ",
    min_freshness=200,
    limit=50
)


compare_personas_tool(
    persona_id1="persona_123",
    persona_id2="persona_456"
)

** :**

  • MBTI

6.

get_statistics_tool()

** :**

  • MBTI

get_persona_survey_history_tool(persona_id="persona_123")

7.

generate_sample_personas_tool(
    count=50,
    use_llm=True  # False     ( )
)


generate_sample_surveys_tool(
    persona_count=20,
    responses_per_persona=2,
    use_llm=True
)


generate_all_sample_data_tool(
    persona_count=50,
    responses_per_persona=2,
    use_llm=True
)

1.

pip install -r requirements.txt

pyproject.toml :

pip install -e .

2.

.env :

# Groq API  (LLM    )
GROQ_API_KEY=your_groq_api_key_here

#  URL (, : sqlite:///data/fgi.db)
DATABASE_URL=sqlite:///data/fgi.db

#   (, : INFO)
LOG_LEVEL=INFO

Groq API :

  1. Groq Console
  2. API Keys
  3. .env

3.

 :

python -m src.main

data/fgi.db ( ).

** : **

    :
  • get_basic_profile_questions_tool() , update_basic_profile_tool()

4.

python -m src.main

MCP

MCP (Cursor AI, Claude Desktop ) .

Cursor AI

  1. **Cursor **

    • Cmd + , (Mac) Ctrl + , (Windows/Linux)
    • "MCP"

  2. **MCP **

    :

    {
  "mcpServers": {
    "persona-server": {
      "command": "python3",
      "args": ["-m", "src.main"],
      "cwd": "/Users/seodong-ug/Desktop/persona_mcp"
    }
  }
}

    : cwd .

  3. **Cursor **

    • Cursor
    • " "

Claude Desktop

claude_desktop_config.json :

{
  "mcpServers": {
    "persona-server": {
      "command": "python3",
      "args": ["-m", "src.main"],
      "cwd": "/Users/seodong-ug/Desktop/persona_mcp"
    }
  }
}

API
health
get_tool_definitions

get_basic_profile_questions_tool
check_basic_profile_tool (persona_id )
update_basic_profile_tool persona_id, profile_data
predict_persona_tool responses
get_persona_tool persona_id
list_personas_tool ( )
compare_personas_tool persona_id1, persona_id2
check_freshness_tool persona_id

FGI
create_fgi_survey_tool category
generate_next_question_tool survey_id, previous_qa
submit_fgi_response_tool persona_id, survey_id, responses
get_persona_survey_history_tool persona_id

analyze_preference_tool query
get_trend_insights_tool topic
search_responses_tool keyword
get_statistics_tool

generate_sample_personas_tool ()
generate_sample_surveys_tool ()
generate_all_sample_data_tool ()

Persona ()

{
    "id": "persona_123",                    #  ID
    "user_id": "user_123",                  #  ID ()
    "freshness_score": 250.0,               #  
    "last_updated": "2024-01-15T10:30:00",  #  
    "created_at": "2024-01-10T09:00:00",    # 
    
    #  
    "age_range": "20 ",
    "gender": "",
    "location": "",
    "occupation_category": "IT/",
    "income_range": "300-500",
    
    #   
    "interests": ["", "AI", ""],
    "personality_traits": ["", " "],
    "values": ["", ""],
    "lifestyle": "    ",
    
    # MBTI 
    "mbti_type": "INTJ",
    "mbti_dimensions": {
        "E_I": "(I) -     ",
        "S_N": "(N) -   ",
        "T_F": "(T) -   ",
        "J_P": "(J) -   "
    },
    "mbti_confidence": 85  # 0-100
}

FGISurvey ()

{
    "id": "survey_123",
    "category": "technology",
    "questions": [
        {
            "id": "q1",
            "text": "     ?",
            "type": "open"
        }
    ],
    "created_at": "2024-01-15T10:00:00",
    "news_source": null  # 
}

FGIResponse ()

{
    "id": "response_123",
    "persona_id": "persona_123",
    "survey_id": "survey_123",
    "responses": [
        {
            "question_id": "q1",
            "answer": "AI   "
        }
    ],
    "submitted_at": "2024-01-15T10:30:00"
}

1:

# 1.     
questions = get_basic_profile_questions_tool()
#  :
# {
#   "questions": {
#     "age_range": {"text": " ", "type": "choice", "options": [...]},
#     "gender": {"text": " ", "type": "choice", "options": [...]},
#     ...
#   }
# }

# 2.     ()
#  UI       .

# 3.     
update_basic_profile_tool(
    persona_id="persona_new_001",
    profile_data={
        "age_range": "20 ",        # questions["questions"]["age_range"]["options"]  
        "gender": "",                 # questions["questions"]["gender"]["options"]  
        "location": "",               # questions["questions"]["location"]["options"]  
        "occupation_category": "IT/"  # questions["questions"]["occupation_category"]["options"]  
        # income_range   
    }
)

# 4.    
survey = create_fgi_survey_tool(
    category="technology",
    persona_id="persona_new_001"
)

# 5.    
submit_fgi_response_tool(
    persona_id="persona_new_001",
    survey_id=survey["survey_id"],
    responses=[
        {
            "question_id": survey["questions"][0]["id"],
            "answer": "AI   .  ChatGPT   AI  ."
        }
    ]
)

# 6.   
next_q = generate_next_question_tool(
    survey_id=survey["survey_id"],
    previous_qa=[
        {
            "question": survey["questions"][0]["text"],
            "answer": "AI   ..."
        }
    ],
    category="technology"
)

# 7.  - ...
# 8.   
persona = predict_persona_tool(
    responses=[
        {"question_id": "q1", "answer": "..."},
        {"question_id": "q2", "answer": "..."}
    ]
)

2:

#    20   
preference = analyze_preference_tool(
    query=" ",
    category="fashion",
    age_range="20",
    gender="",
    location="",
    limit=30
)

#   
print(preference["insights"])
# - "   "
# - "  "
# - "   "

3:

#   
trends = get_trend_insights_tool(
    topic="",
    persona_filters={
        "age_range": "20",
        "gender": ""
    },
    min_freshness=150
)

#   
print(trends["summary"])

4:

# 50    (LLM )
generate_sample_personas_tool(count=50, use_llm=True)

# 20    2   
generate_sample_surveys_tool(
    persona_count=20,
    responses_per_persona=2,
    use_llm=True
)

#       
generate_all_sample_data_tool(
    persona_count=50,
    responses_per_persona=2,
    use_llm=False  #     ( )
)

5:

#   
comparison = compare_personas_tool(
    persona_id1="persona_123",
    persona_id2="persona_456"
)

#   
print(comparison["similarities"])
print(comparison["differences"])

#   
stats = get_statistics_tool()
print(f"  : {stats['total_personas']}")
print(f" : {stats['avg_freshness']}")
print(f"MBTI : {stats['mbti_distribution']}")

  • FastMCP: MCP
  • SQLAlchemy: ORM
  • Groq API: LLM
  • Pydantic:
  • Python-dotenv:
  • SQLite: (PostgreSQL )

persona_mcp/
 src/
    main.py                 # MCP   
    db/
       database.py        #     
       models.py          # SQLAlchemy  
    tools/
       persona/
           fgi_tools.py   # FGI   
           predict.py     #  
           freshness.py   #  
           insights.py    #  
           query.py       #  
           sample_data.py #   
    utils/
        cache.py           #  
        credentials.py     # API  
 data/
    fgi.db                 # SQLite 
 cache/                     #   
 requirements.txt           # Python 
 pyproject.toml            #  
 README.md                 #

  1. **Groq API **: LLM Groq API . CRUD , .

  2. ** **: . .

  3. ****: , 12 . .

  4. ****: SQLite , DATABASE_URL PostgreSQL, MySQL .

 .

!

 .
查看源码 项目主页

相关 MCP 服务

思维导图 MCP
WEIAIb/mind-map-mcp
一个用于生成思维导图的 MCP (Model Context Protocol) 服务器。 看起来您的请求中没有提供具体的英文技术文档内容以供翻译。如果您能分享更多详细信息或具体段落，我将能够帮助您将其准确地翻译成中文，同时保留代码块、链接、格式结构和专业术语的准确性。请提供需要翻译的具体文本。
热闻社媒服务
@wopal-cn/mcp-hotnews-server
一个模型上下文协议服务器，提供来自主要中文社交平台和新闻网站的实时热门话题。
搞
搞定ppt
@NeekChaw/okppt
此项目是让大型语言模型（如Claude、GPT等）能够自主设计PowerPoint演示文稿的"曲线救国"解决方案。通过让AI生成SVG图像，再借助本工具将其全屏插入PPT幻灯片，我们成功实现了AI完全控制PPT设计的能力，而无需直接操作复杂的PPT对象模型。
快递100 MCP Server
kuaidi100/kuaidi100-mcp
快递100 MCP Server提供快递信息查询、快递运费预估比价、快递时效查询（含发货前时效查询与在途动态时效查询）等功能。