mcp

AI-native access to PS-Wiki terminology via Model Context Protocol

The PS-Wiki MCP Server provides AI assistants with direct access to power systems terminology through the Model Context Protocol. This enables natural language queries with accurate, cited responses.

Features

  • 🔍 Search 173+ power systems terms by keyword or concept
  • 📖 Retrieve detailed definitions with equations and citations
  • 🔗 Explore related concepts and build knowledge graphs
  • 🏷️ Browse by tags (stability, control, protection, market, etc.)
  • 🤖 Works with Claude Desktop, Ollama, and other MCP clients

Quick Start

Installation

# Install from PyPI (when published)
pip install pswiki-mcp

# Or install from source
git clone https://github.com/ps-wiki/ps-wiki.github.io.git
cd ps-wiki.github.io/mcp
pip install -e .

Configuration

Option 1: Claude Desktop (Easiest)

Add to claude_desktop_config.json:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "pswiki": {
      "command": "python",
      "args": ["-m", "pswiki_mcp"]
    }
  }
}

Restart Claude Desktop, then ask:

  • “Search for power systems stability terms”
  • “Explain voltage stability in detail”
  • “What are related concepts to automatic generation control?”

Option 2: Ollama (Open-Source)

For local, private AI using Ollama:

Using Continue.dev (VS Code)

  1. Install Continue extension
  2. Configure ~/.continue/config.json:
{
  "models": [
    {
      "title": "Llama 3.2",
      "provider": "ollama",
      "model": "llama3.2"
    }
  ],
  "mcpServers": {
    "pswiki": {
      "command": "python",
      "args": ["-m", "pswiki_mcp"]
    }
  }
}
  1. Open Continue sidebar (⌘+L) and start asking questions!

Quick Test Script

cd ps-wiki.github.io/mcp

# Make sure Ollama is running
ollama pull llama3.2

# Run demo
python quick_test.py "voltage stability"

This demonstrates the complete workflow: MCP fetches data, Ollama explains it.

Available Tools

The MCP server provides 5 tools for AI assistants:

search_terms

Search terminology by keyword, phrase, or concept.

Example: search_terms(query="frequency control", limit=10)

get_term

Retrieve full details for a specific term.

Example: get_term(term_id="voltage-stability")

Find related concepts with configurable depth.

Example: get_related_terms(term_id="power-flow", depth=2)

list_tags

List all available tags with usage counts.

Example: list_tags()

get_terms_by_tag

Filter terms by category.

Example: get_terms_by_tag(tag="stability")

Usage Examples

With Claude Desktop

You: “Explain the difference between voltage stability and frequency stability”

Claude: [Uses search_terms and get_term tools to fetch PS-Wiki data, then provides accurate explanation with citations]

With Ollama

You: “What is automatic generation control?”

AI: [Fetches AGC definition from PS-Wiki via MCP, explains using local Llama model]

Advantage: Complete privacy - all processing happens on your machine.

Programmatic Usage

from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client

server_params = StdioServerParameters(
    command="python",
    args=["-m", "pswiki_mcp"]
)

async with stdio_client(server_params) as (read, write):
    async with ClientSession(read, write) as session:
        await session.initialize()
        result = await session.call_tool(
            "search_terms",
            {"query": "stability", "limit": 5}
        )

Data Source

The MCP server fetches data from the PS-Wiki REST API hosted on Cloudflare Workers. All data is read-only and licensed under CC BY-NC 4.0.

Benefits

✅ AI-Native: Natural language access to technical terminology
✅ Accurate: Cited definitions from standards and papers
✅ Private: Use with Ollama for complete local processing
✅ Free: No API costs, no rate limits
✅ Extensible: Works with any MCP-compatible client

Resources

Support