DynamoDB MCP Server

A Model Context Protocol server for managing Amazon DynamoDB resources. This server provides tools for table management, capacity management, and data operations.

Author

Iman Kamyabi (ikmyb@icloud.com)

Features

Table Management

• Create new DynamoDB tables with customizable configurations
• List existing tables
• Get detailed table information
• Configure table settings

Index Management

• Create and manage Global Secondary Indexes (GSI)
• Update GSI capacity
• Create Local Secondary Indexes (LSI)

Capacity Management

• Update provisioned read/write capacity units
• Manage table throughput settings

Data Operations

• Insert or replace items in tables
• Retrieve items by primary key
• Update specific item attributes
• Query tables with conditions
• Scan tables with filters

Neo4j Action Tracking

• Record all MCP actions in a Neo4j graph database
• Find patterns in action sequences
• Get recommendations based on historical data
• Track relationships between different MCPs
• Build a knowledge graph of team actions

Note: Delete operations are not supported to prevent accidental data loss.

Setup

1. Install dependencies:

npm install

2. Configure AWS credentials as environment variables:

export AWS_ACCESS_KEY_ID="your_access_key"
export AWS_SECRET_ACCESS_KEY="your_secret_key"
export AWS_REGION="your_region"

3. Configure Neo4j connection (required for action tracking):

export NEO4J_URI="bolt://localhost:7687"
export NEO4J_USERNAME="neo4j"
export NEO4J_PASSWORD="your_password"

4. Initialize Neo4j schema (first-time setup):

npm run init-neo4j

5. Build the server:

npm run build

6. Start the server:

npm start

7. Enable Docker container (optional):

docker build -t mcp/dynamodb-mcp-server -f Dockerfile .
docker run -p 3000:3000 -e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY -e AWS_REGION -e NEO4J_URI -e NEO4J_USERNAME -e NEO4J_PASSWORD mcp/dynamodb-mcp-server

Configuration

Environment Variables

• AWS_ACCESS_KEY_ID: Your AWS access key
• AWS_SECRET_ACCESS_KEY: Your AWS secret access key
• AWS_REGION: Your AWS region (e.g., us-east-1)
• AWS_SESSION_TOKEN: (Optional) Session token for temporary credentials
• NEO4J_URI: URI for Neo4j database connection
• NEO4J_USERNAME: Username for Neo4j database
• NEO4J_PASSWORD: Password for Neo4j database
• CURSOR_USER_ID: (Optional) Default user ID for Cursor integration
• CURSOR_USER_EMAIL: (Optional) Default user email for Cursor integration
• CURSOR_USER_NAME: (Optional) Default user name for Cursor integration
• CURSOR_USER_TEAM: (Optional) Default user team for Cursor integration

Neo4j Setup

For optimal performance, create the following indexes in your Neo4j database:

CREATE CONSTRAINT IF NOT EXISTS FOR (user:User) REQUIRE user.id IS UNIQUE;
CREATE CONSTRAINT IF NOT EXISTS FOR (mcp:MCP) REQUIRE mcp.id IS UNIQUE;
CREATE CONSTRAINT IF NOT EXISTS FOR (action:Action) REQUIRE action.id IS UNIQUE;
CREATE INDEX IF NOT EXISTS FOR (user:User) ON (user.email);
CREATE INDEX IF NOT EXISTS FOR (user:User) ON (user.team);
CREATE INDEX IF NOT EXISTS FOR (action:Action) ON (action.type);
CREATE INDEX IF NOT EXISTS FOR (action:Action) ON (action.name);
CREATE FULLTEXT INDEX actionContext IF NOT EXISTS FOR (a:Action) ON EACH [a.name, a.type];

The fulltext index enables context-based action recommendations.

Tools

create_table

Creates a new DynamoDB table with specified configuration.

Parameters:

• tableName: Name of the table to create
• partitionKey: Name of the partition key
• partitionKeyType: Type of partition key (S=String, N=Number, B=Binary)
• sortKey: (Optional) Name of the sort key
• sortKeyType: (Optional) Type of sort key
• readCapacity: Provisioned read capacity units
• writeCapacity: Provisioned write capacity units

Example:

{
  "tableName": "Users",
  "partitionKey": "userId",
  "partitionKeyType": "S",
  "readCapacity": 5,
  "writeCapacity": 5
}

list_tables

Lists all DynamoDB tables in the account.

Parameters:

• limit: (Optional) Maximum number of tables to return
• exclusiveStartTableName: (Optional) Name of the table to start from for pagination

Example:

{
  "limit": 10
}

describe_table

Gets detailed information about a DynamoDB table.

Parameters:

• tableName: Name of the table to describe

Example:

{
  "tableName": "Users"
}

create_gsi

Creates a global secondary index on a table.

Parameters:

• tableName: Name of the table
• indexName: Name of the new index
• partitionKey: Partition key for the index
• partitionKeyType: Type of partition key
• sortKey: (Optional) Sort key for the index
• sortKeyType: (Optional) Type of sort key
• projectionType: Type of projection (ALL, KEYS_ONLY, INCLUDE)
• nonKeyAttributes: (Optional) Non-key attributes to project
• readCapacity: Provisioned read capacity units
• writeCapacity: Provisioned write capacity units

Example:

{
  "tableName": "Users",
  "indexName": "EmailIndex",
  "partitionKey": "email",
  "partitionKeyType": "S",
  "projectionType": "ALL",
  "readCapacity": 5,
  "writeCapacity": 5
}

update_gsi

Updates the provisioned capacity of a global secondary index.

Parameters:

• tableName: Name of the table
• indexName: Name of the index to update
• readCapacity: New read capacity units
• writeCapacity: New write capacity units

Example:

{
  "tableName": "Users",
  "indexName": "EmailIndex",
  "readCapacity": 10,
  "writeCapacity": 10
}

create_lsi

Creates a local secondary index on a table (must be done during table creation).

Parameters:

• tableName: Name of the table
• indexName: Name of the new index
• partitionKey: Partition key for the table
• partitionKeyType: Type of partition key
• sortKey: Sort key for the index
• sortKeyType: Type of sort key
• projectionType: Type of projection (ALL, KEYS_ONLY, INCLUDE)
• nonKeyAttributes: (Optional) Non-key attributes to project
• readCapacity: (Optional) Provisioned read capacity units
• writeCapacity: (Optional) Provisioned write capacity units

Example:

{
  "tableName": "Users",
  "indexName": "CreatedAtIndex",
  "partitionKey": "userId",
  "partitionKeyType": "S",
  "sortKey": "createdAt",
  "sortKeyType": "N",
  "projectionType": "ALL"
}

update_capacity

Updates the provisioned capacity of a table.

Parameters:

• tableName: Name of the table
• readCapacity: New read capacity units
• writeCapacity: New write capacity units

Example:

{
  "tableName": "Users",
  "readCapacity": 10,
  "writeCapacity": 10
}

put_item

Inserts or replaces an item in a table.

Parameters:

• tableName: Name of the table
• item: Item to put into the table (as JSON object)

Example:

{
  "tableName": "Users",
  "item": {
    "userId": "123",
    "name": "John Doe",
    "email": "john@example.com"
  }
}

get_item

Retrieves an item from a table by its primary key.

Parameters:

• tableName: Name of the table
• key: Primary key of the item to retrieve

Example:

{
  "tableName": "Users",
  "key": {
    "userId": "123"
  }
}

update_item

Updates specific attributes of an item in a table.

Parameters:

• tableName: Name of the table
• key: Primary key of the item to update
• updateExpression: Update expression
• expressionAttributeNames: Attribute name mappings
• expressionAttributeValues: Values for the update expression
• conditionExpression: (Optional) Condition for update
• returnValues: (Optional) What values to return

Example:

{
  "tableName": "Users",
  "key": {
    "userId": "123"
  },
  "updateExpression": "SET #n = :name",
  "expressionAttributeNames": {
    "#n": "name"
  },
  "expressionAttributeValues": {
    ":name": "Jane Doe"
  }
}

query_table

Queries a table using key conditions and optional filters.

Parameters:

• tableName: Name of the table
• keyConditionExpression: Key condition expression
• expressionAttributeValues: Values for the key condition expression
• expressionAttributeNames: (Optional) Attribute name mappings
• filterExpression: (Optional) Filter expression for results
• limit: (Optional) Maximum number of items to return

Example:

{
  "tableName": "Users",
  "keyConditionExpression": "userId = :id",
  "expressionAttributeValues": {
    ":id": "123"
  }
}

scan_table

Scans an entire table with optional filters.

Parameters:

• tableName: Name of the table
• filterExpression: (Optional) Filter expression
• expressionAttributeValues: (Optional) Values for the filter expression
• expressionAttributeNames: (Optional) Attribute name mappings
• limit: (Optional) Maximum number of items to return

Example:

{
  "tableName": "Users",
  "filterExpression": "age > :minAge",
  "expressionAttributeValues": {
    ":minAge": 21
  }
}

upassistant_get_item_by_id

Retrieves an UpAssistant item by id from the correct table based on environment (uat/prod).

Parameters:

• id: ID of the item to retrieve
• env: Environment: ‘uat’ or ‘prod’

Example:

{
  "id": "97729d8e-b722-4822-9490-a900cec81260",
  "env": "prod"
}

upassistant_put_item

Puts an UpAssistant item into the correct table based on environment (uat/prod).

Parameters:

• item: Item to put into the table (as JSON object)
• env: Environment: ‘uat’ or ‘prod’

Example:

{
  "item": {
    "id": "97729d8e-b722-4822-9490-a900cec81260",
    "createdAt": "2024-10-29T10:22:47.109375",
    "description": "Birlikte etkin dinleme çalışması yapalım mı?",
    "extra": {},
    "frequencyPenalty": "0",
    "introductionMessages": [
      {
        "type": "default",
        "value": "Merhaba! nnBirlikte deneme yapmadan önce işe yaradığını gördüğüm birkaç ipucu paylaşayım: nn- Karşındakine tam dikkatini ver - telefonu bir kenara bırak! n- Sadece sözleri değil, beden dilini de okun- Sözünü kesme, sabırla dinlen- "Seni anlıyorum" demek yerine, duyduklarını özetlen- Merak et ve soru sor - ama sorgulamak için değil, anlamak için!n- Empati kur - "Ben olsam ne hissederdim?" diye düşünnnBöyle sohbetler daha keyifli ve anlamlı oluyor.. nnNe dersin, bir deneyelim mi? "
      },
      {
        "type": "user-input",
        "value": "Konuşma kiminle olacak? [BLANK]Konuşma ne hakkında olacak? [BLANK]"
      }
    ],
    "maxTokens": "800",
    "modelName": "GPT-4o",
    "name": "Etkin Dinleme",
    "presencePenalty": "1.0",
    "prompt": "",
    "src": "https://upwagmidevcontent234355-upwagmitec.s3.us-east-1.amazonaws.com/public/up_app_gorseller/Etkin+dinleme.jpeg",
    "status": true,
    "temperature": "0.9",
    "template": [
      {
        "key": "instructions",
        "title": "Instructions:",
        "type": "name-value-list",
        "value": []
      },
      {
        "key": "additionalConsideration",
        "title": "Additional Consideration:",
        "type": "name-value-list",
        "value": []
      }
    ],
    "title": "Birlikte etkin dinleme çalışması yapalım mı?",
    "topP": "0.95",
    "type": "user-input",
    "updatedAt": "2024-12-25T09:39:42.272974",
    "userId": "7e30775e-cbfc-4fb1-8d4b-7bac7e7210af"
  },
  "env": "prod"
}

Sample Questions

Here are some example questions you can ask Claude when using this DynamoDB MCP server:

Table Management

• “Create a new DynamoDB table called ‘Products’ with a partition key ‘productId’ (string) and sort key ‘timestamp’ (number)”
• “List all DynamoDB tables in my account”
• “What’s the current configuration of the Users table?”
• “Add a global secondary index on the email field of the Users table”

Capacity Management

• “Update the Users table capacity to 20 read units and 15 write units”
• “Scale up the EmailIndex GSI capacity on the Users table”
• “What’s the current provisioned capacity for the Orders table?”

Data Operations

• “Insert a new user with ID ‘123’, name ‘John Doe’, and email ‘john@example.com’”
• “Get the user with ID ‘123’”
• “Update the email address for user ‘123’ to ‘john.doe@example.com’”
• “Find all orders placed by user ‘123’”
• “List all users who are over 21 years old”
• “Query the EmailIndex to find the user with email ‘john@example.com’”

Building

Docker:

docker build -t mcp/dynamodb-mcp-server -f Dockerfile .

Usage with Cursor Desktop

Add this to your cursor_desktop_config.json:

{
  "mcpServers": {
    "dynamodb": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "AWS_ACCESS_KEY_ID",
        "-e",
        "AWS_SECRET_ACCESS_KEY",
        "-e",
        "AWS_REGION",
        "-e",
        "NEO4J_URI",
        "-e",
        "NEO4J_USERNAME",
        "-e",
        "NEO4J_PASSWORD"
      ],
      "env": {
        "AWS_ACCESS_KEY_ID": "your_access_key",
        "AWS_SECRET_ACCESS_KEY": "your_secret_key",
        "AWS_REGION": "your_region",
        "NEO4J_URI": "bolt://your-neo4j-instance:7687",
        "NEO4J_USERNAME": "neo4j",
        "NEO4J_PASSWORD": "your_password"
      },
      "includeAuth": true,
      "authHeaders": {
        "cursor-auth": "${BASE64_ENCODED_USER_INFO}"
      }
    }
  }
}

The cursor-auth header will be automatically populated with the current user’s information when you configure "includeAuth": true.

Development

To run in development mode with auto-reloading:

npm run dev

Neo4j Action Tracking

This MCP server includes integrated Neo4j action tracking to record and analyze MCP actions across different services. This helps in:

1. Recording all actions taken through MCPs
2. Finding patterns in action sequences
3. Suggesting next actions based on historical data
4. Recommending actions based on context
5. Tracking user identity and team membership
6. Building organizational memory and knowledge retention

Cursor Integration

The MCP server now includes full integration with Cursor, allowing actions to be attributed to specific users:

{
  "mcpServers": {
    "dynamodb": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "AWS_ACCESS_KEY_ID",
        "-e",
        "AWS_SECRET_ACCESS_KEY",
        "-e",
        "AWS_REGION",
        "-e",
        "NEO4J_URI",
        "-e",
        "NEO4J_USERNAME",
        "-e",
        "NEO4J_PASSWORD",
        "mcp/dynamodb-mcp-server"
      ],
      "env": {
        "AWS_ACCESS_KEY_ID": "your_access_key",
        "AWS_SECRET_ACCESS_KEY": "your_secret_key",
        "AWS_REGION": "your_region",
        "NEO4J_URI": "bolt://your-neo4j-instance:7687",
        "NEO4J_USERNAME": "neo4j",
        "NEO4J_PASSWORD": "your_password"
      },
      "includeAuth": true,
      "authHeaders": {
        "cursor-auth": "<BASE64_ENCODED_USER_INFO>"
      }
    }
  }
}

The cursor-auth header should contain base64-encoded JSON with user identity information:

{
  "userId": "unique_user_id",
  "userName": "User's Display Name",
  "userEmail": "user@example.com",
  "userTeam": "Engineering"
}

User Identity Management

The Neo4j action tracking system now includes enhanced user identity features:

• User Nodes: Tracked in Neo4j with attributes like name, email, and team
• Identity Preservation: Users are consistently identified across sessions
• Team Analysis: Actions can be analyzed at both individual and team levels
• Organization Insights: Build understanding of how different teams interact with resources

Action Tracking Tools

record_action

Records an MCP action in the Neo4j graph database.

Parameters:

• userId: ID of the user performing the action
• userName: Name of the user
• mcpId: ID of the MCP being used
• mcpType: Type of MCP (DynamoDB, Mattermost, Jira, etc.)
• mcpName: Name of the MCP instance
• actionType: Type of action being performed
• actionName: Specific action name
• parameters: Action parameters
• result: Action result
• status: Status of the action (“success” or “failure”)

Example:

{
  "userId": "user123",
  "userName": "John Doe",
  "mcpId": "mattermost-mcp",
  "mcpType": "Mattermost",
  "mcpName": "Team Chat",
  "actionType": "message",
  "actionName": "post_message",
  "parameters": {
    "channelId": "general",
    "message": "Hello team!"
  },
  "result": {
    "messageId": "abc123"
  },
  "status": "success"
}

get_similar_actions

Finds similar actions to the current one.

Parameters:

• mcpType: Type of MCP
• actionType: Type of action
• parameters: Parameters of the action
• limit: (Optional) Maximum number of similar actions to return

Example:

{
  "mcpType": "DynamoDB",
  "actionType": "table_management",
  "parameters": {
    "tableName": "Users"
  }
}

get_user_history

Gets a user’s action history.

Parameters:

• userId: ID of the user
• limit: (Optional) Maximum number of actions to return

Example:

{
  "userId": "user123",
  "limit": 10
}

suggest_next_action

Suggests the next action based on typical patterns.

Parameters:

• userId: ID of the user
• mcpType: Type of MCP
• currentActionType: Type of the current action
• currentParameters: Parameters of the current action

Example:

{
  "userId": "user123",
  "mcpType": "DynamoDB",
  "currentActionType": "create_table",
  "currentParameters": {
    "tableName": "Users"
  }
}

get_action_recommendations

Gets action recommendations based on a context description.

Parameters:

• userId: ID of the user
• context: Context description to find relevant actions

Example:

{
  "userId": "user123",
  "context": "setting up a new user authentication system"
}

Benefits of Action Tracking

The Neo4j action tracking system provides several key benefits:

1. Knowledge Retention: Captures organizational knowledge that would otherwise be lost
2. Pattern Recognition: Identifies common workflows and successful approaches
3. Smart Recommendations: Suggests relevant actions based on historical patterns
4. Cross-Service Insights: Shows dependencies between different systems
5. Team Collaboration: Enhances knowledge sharing between team members
6. Workflow Optimization: Identifies inefficient patterns and bottlenecks
7. Time Savings: Reduces rediscovery of solutions and minimizes duplicated efforts

As your team uses the system more, the Neo4j knowledge graph becomes increasingly valuable, providing deeper insights and more accurate recommendations.

Cross-MCP Integration

The Neo4j Action Tracking system is designed to work across multiple MCPs including:

• DynamoDB MCP (this server)
• Mattermost MCP
• Jira MCP
• And other future MCPs

This integration allows your team to build a comprehensive knowledge graph of actions taken across different systems, enabling smart recommendations and workflow optimizations.

License

This MCP server is licensed under the MIT License. This means you are free to use, modify, and distribute the software, subject to the terms and conditions of the MIT License. For more details, please see the LICENSE file in the project repository.