Build a Knowledge Base MCP Server

Objective

Build a Python class that simulates an MCP knowledge base server with search, document retrieval, and document addition capabilities.

Requirements

Create a class KnowledgeBase that:

1. Stores documents in an internal dictionary:

   • Documents have: id, title, content, tags

2. Lists available resources with list_resources() -> list[dict]:

   • Returns list of resources for each document
   • Each resource has: uri, name, description

3. Reads a resource with read_resource(uri: str) -> str:

   • Parses document ID from URI (format: doc://id)
   • Returns document content
   • Raises ValueError if not found

4. Defines tools with get_tool_definitions() -> list[dict]:

   • search_documents: Search by query string
   • add_document: Add new document

5. Executes tools with call_tool(name: str, arguments: dict) -> dict:

   • search_documents: Returns matching documents (title contains query)
   • add_document: Adds document, returns new ID

Example Usage

kb = KnowledgeBase()

# Add documents
result = kb.call_tool("add_document", {
    "title": "Python Basics",
    "content": "Python is a programming language...",
    "tags": ["python", "programming"]
})
# {"success": True, "id": "doc_1"}

# Search documents
result = kb.call_tool("search_documents", {"query": "Python"})
# {"results": [{"id": "doc_1", "title": "Python Basics", ...}]}

# List resources
resources = kb.list_resources()
# [{"uri": "doc://doc_1", "name": "Python Basics", ...}]

# Read resource
content = kb.read_resource("doc://doc_1")
# "Python is a programming language..."

Hints

• Use a counter for generating unique IDs (doc_1, doc_2, etc.)
• Search should be case-insensitive
• Resources should be dynamically generated from stored documents