Skip to main content
Glama
dannwaneri

MCP Knowledge Base Server

by dannwaneri
README.md
\# MCP Knowledge Base Server



A Model Context Protocol (MCP) server that exposes a knowledge base through multiple tools for semantic search and data retrieval.



\## Features



\- \*\*search\_knowledge\*\*: Search for relevant information using keywords

\- \*\*list\_categories\*\*: List all available categories

\- \*\*get\_by\_category\*\*: Filter entries by category

\- \*\*get\_by\_id\*\*: Retrieve specific entries by ID

\- \*\*advanced\_search\*\*: Search with category filters and result limits

\- \*\*In-memory caching\*\*: 60-second TTL for faster repeated queries



\## Setup

```bash

npm install

npm run build

```



\## Usage with Claude Desktop



Add to your Claude Desktop config (`%APPDATA%\\Claude\\claude\_desktop\_config.json`):

```json

{

  "mcpServers": {

    "knowledge-base": {

      "command": "node",

      "args": \["C:\\\\path\\\\to\\\\mcp-learning\\\\dist\\\\index.js"]

    }

  }

}

```



\## Tech Stack



\- TypeScript

\- @modelcontextprotocol/sdk

\- Node.js



\## Next Steps



\- \[ ] Connect to Vectorize for semantic search

\- \[ ] Deploy to Cloudflare Workers

\- \[ ] Add authentication

TDQS

B3.3/5.0

Scored across 5 tools

Disambiguation3/5

There is some overlap between 'advanced_search' and 'search_knowledge', as both are for searching, though 'advanced_search' adds filters. 'get_by_category' and 'list_categories' are distinct but related. The descriptions help differentiate, but an agent might initially confuse the two search tools.

Naming Consistency4/5

Most tools follow a clear verb_noun pattern (e.g., 'get_by_id', 'list_categories'), with 'advanced_search' and 'search_knowledge' being slight deviations. The naming is mostly consistent and readable, with only minor inconsistencies in verb usage.

Tool Count5/5

With 5 tools, this server is well-scoped for a knowledge base domain. Each tool serves a distinct purpose (searching, retrieving by ID/category, listing categories), and the count is appropriate without being too sparse or overwhelming.

Completeness4/5

The toolset covers core CRUD-like operations for a knowledge base: searching, retrieving by ID and category, and listing categories. A minor gap is the lack of create, update, or delete tools, but for a read-only knowledge base server, this is reasonable and agents can work around it.

Maintenance

ActivityInactive
ResponsivenessNo issues