MCP Tools

Distributed Knowledge exposes its functionality through a set of Model Context Protocol (MCP) tools. This document provides a detailed reference for these tools, including their purpose, parameters, and usage examples.

Query Management Tools

These tools handle the creation, tracking, and management of queries in the network.

cqAskQuestion

Sends a question to specified peers or broadcasts it to the entire network.

Parameters:

question (string, required): The text of the question to send
peers (array of strings, required): List of peer identifiers to receive the question; leave empty to broadcast to all peers

Example:

{
  "name": "cqAskQuestion",
  "parameters": {
    "question": "What are the latest developments in quantum computing?",
    "peers": ["quantum_expert", "physics_researcher"]
  }
}

Response:

{
  "content": "Query request sent ... Instruct the user to ask the model for summarize on the query What are the latest developments in quantum computing?"
}

cqListRequestedQueries

Retrieves all requested queries, optionally filtered by status or sender.

Parameters:

status (string, optional): Status filter (e.g., 'pending', 'accepted', 'rejected')
from (string, optional): Sender filter (peer identifier)

Example:

{
  "name": "cqListRequestedQueries",
  "parameters": {
    "status": "pending"
  }
}

Response:

{
  "qry-123": {
    "id": "qry-123",
    "from": "user1",
    "question": "What are the latest developments in quantum computing?",
    "status": "pending",
    "timestamp": "2025-01-15T10:30:45Z"
  },
  "qry-124": {
    "id": "qry-124",
    "from": "user2",
    "question": "How do neural networks work?",
    "status": "pending",
    "timestamp": "2025-01-15T11:15:22Z"
  }
}

cqSummarizeAnswers

Retrieves all peer responses for a given question and returns a cohesive summary.

Parameters:

related_question (string, required): The question for which to fetch and analyze responses
detailed_answer (number, optional): Set to 1 for detailed response, 0 for concise summary

Example:

{
  "name": "cqSummarizeAnswers",
  "parameters": {
    "related_question": "What are the latest developments in quantum computing?",
    "detailed_answer": 1
  }
}

Response: A comprehensive summary of all answers received from network peers.

cqUpdateEditAnswer

Edits the content of a specific answer.

Parameters:

query_id (string, required): ID of the query to update
new_answer (string, required): New answer content

Example:

{
  "name": "cqUpdateEditAnswer",
  "parameters": {
    "query_id": "qry-123",
    "new_answer": "Recent developments in quantum computing include improvements in qubit stability and advances in quantum error correction..."
  }
}

Response:

{
  "content": "Successfully updated answer for query_id 'qry-123'."
}

Approval Management Tools

These tools manage the automatic and manual approval of queries and responses.

cqAddAutoApprovalCondition

Adds a condition to the automatic approval system.

Parameters:

sentence (string, required): The condition to add

Example:

{
  "name": "cqAddAutoApprovalCondition",
  "parameters": {
    "sentence": "Allow questions about scientific topics from academic users"
  }
}

Response:

{
  "content": "Condition added successfully: Allow questions about scientific topics from academic users"
}

cqRemoveAutoApprovalCondition

Removes a condition from the automatic approval system.

Parameters:

condition (string, required): The exact text of the condition to remove

Example:

{
  "name": "cqRemoveAutoApprovalCondition",
  "parameters": {
    "condition": "Allow questions about scientific topics from academic users"
  }
}

Response:

{
  "content": "Condition 'Allow questions about scientific topics from academic users' removed successfully."
}

cqListAutoApprovalConditions

Lists all conditions in the automatic approval system.

Parameters: None

Example:

{
  "name": "cqListAutoApprovalConditions"
}

Response:

[
  "Allow questions about scientific topics from academic users",
  "Accept queries related to programming from any user",
  "Approve all questions from verified_researcher"
]

cqAcceptQuery

Marks a pending query as 'accepted' and sends the answer to the requester.

Parameters:

id (string, required): Unique identifier of the query to accept

Example:

{
  "name": "cqAcceptQuery",
  "parameters": {
    "id": "qry-123"
  }
}

Response:

{
  "content": "Question 'What are the latest developments in quantum computing?' has been accepted."
}

cqRejectQuery

Marks a pending query as 'rejected' and notifies the requester.

Parameters:

id (string, required): Unique identifier of the query to reject

Example:

{
  "name": "cqRejectQuery",
  "parameters": {
    "id": "qry-123"
  }
}

Response:

{
  "content": "Question 'What are the latest developments in quantum computing?' has been rejected."
}

Knowledge Management Tools

These tools manage the knowledge base used by the RAG system.

updateKnowledgeSources

Updates knowledge sources by adding new content to the RAG database.

Parameters:

file_name (string, optional): Name of the file to add
file_content (string, optional): Content of the file
file_path (string, optional): Path to an existing file

Example using file content:

{
  "name": "updateKnowledgeSources",
  "parameters": {
    "file_name": "quantum_computing.txt",
    "file_content": "Quantum computing is an emerging field that leverages quantum mechanics to perform computations..."
  }
}

Example using file path:

{
  "name": "updateKnowledgeSources",
  "parameters": {
    "file_path": "/path/to/quantum_research.pdf"
  }
}

Response:

{
  "content": "RAG resource 'quantum_computing.txt' added successfully and vector database refreshed."
}

User Management Tools

These tools manage and interact with users in the network.

cqGetActiveUsers

Retrieves lists of active and inactive users from the server.

Parameters:

flag (boolean, optional): Additional options flag

Example:

{
  "name": "cqGetActiveUsers",
  "parameters": {
    "flag": false
  }
}

Response:

{
  "active": ["alice", "bob", "research_team"],
  "inactive": ["charlie", "data_scientist"]
}

cqGetUserDescriptions

Retrieves descriptions associated with a specific user.

Parameters:

user_id (string, required): ID of the user whose descriptions are requested

Example:

{
  "name": "cqGetUserDescriptions",
  "parameters": {
    "user_id": "research_team"
  }
}

Response:

[
  "Quantum physics research group",
  "Specializes in quantum computing theory",
  "Active member since 2023"
]

Best Practices for Using MCP Tools

Tool Sequencing: Use tools in logical sequences for complex operations
Check active users before sending targeted questions
List queries before approving or rejecting them
Add knowledge sources before asking related questions
Error Handling: Be prepared to handle potential errors
Check responses for error messages
Retry operations that might have failed due to temporary issues
Verify the success of critical operations
Parameter Validation: Ensure parameters are correctly formatted
Provide all required parameters
Use correct data types for each parameter
Validate input before sending to avoid errors
Performance Considerations:
Batch operations when possible to reduce roundtrips
Limit the size of knowledge sources to reasonable chunks
Use filters when listing queries to reduce data transfer
Security Awareness:
Only add trustworthy sources to the knowledge base
Review automatic approval conditions regularly
Be cautious about who can send queries to your node