home / mcp / ifs cloud mcp server

IFS Cloud MCP Server

AI-powered MCP server for IFS Cloud codebase analysis, versioned data, and hybrid search capabilities.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "graknol-ifs-cloud-core-mcp-server": {
      "command": "uv",
      "args": [
        "run",
        "python",
        "-m",
        "src.ifs_cloud_mcp_server.main",
        "server",
        "--version",
        "25.1.0"
      ]
    }
  }
}

You run an MCP server that analyzes, indexes, and searches IFS Cloud codebases. It generates dependency maps, PageRank-based importance scores, and semantic/lexical search results to help you understand and navigate large IFS Cloud projects. You can import versions, analyze code, build embeddings, and run a responsive server that serves AI-assisted insights to your preferred MCP client.

How to use

You use an MCP client to interact with the server. First, import your IFS Cloud version, then generate analysis data, start the server, and query for insights. The server supports hybrid search that combines semantic understanding with exact lexical matches, enabling you to quickly locate APIs, business rules, and architectural patterns.

Typical usage pattern you can follow: import a version, run analysis, calculate PageRank, and start the server for interactive exploration. Your AI agent can then perform semantic, lexical, or hybrid searches against the codebase and return ranked results that reflect both meaning and importance within the dependency graph.

When you want to test or demonstrate, you can start by importing a ZIP file, running a full analysis, computing PageRank scores, and then launching the MCP server for immediate querying. Your client can issue semantic queries, exact API lookups, or combined hybrid searches. The system emphasizes version isolation so that each IFS Cloud release has its own analysis and indexes.

How to install

Prerequisites you should have installed on your system: Python, a compatible runtime, and a command runner for managing processes. You will also need a ZIP file of the IFS Cloud version you want to analyze.

# Step 1: Clone the MCP server repository
git clone https://github.com/graknol/ifs-cloud-core-mcp-server.git
cd ifs-cloud-core-mcp-server

# Step 2: Install required dependencies (Python-based MCP server)
uv sync

# If you prefer Python packaging
pip install -e .

Additional notes and steps you may perform

Import an IFS Cloud ZIP file to create a versioned catalog. The import command accepts a ZIP file and a version tag.

Analyze your codebase to extract dependencies, API calls, and other code relationships. This prepares the data needed for PageRank and reference graphs.

Start the MCP server to begin serving AI-assisted searches and analysis outputs to your MCP client. You can also start a web UI if available for interactive exploration.

Supported files and workflows

The MCP server analyzes multiple IFS Cloud file types, including PL/SQL, data entities, UI clients, data projections, and full-stack fragments. It builds a reference graph of file-to-file dependencies and computes PageRank-based importance scores to highlight foundational components and commonly used logic.

Search capabilities overview

You have three main search tools: semantic search using embeddings, lexical search for exact API names and keywords, and a hybrid search that fuses both. CUDA acceleration is available for faster semantic searches, and a neural reranking process (FlashRank) improves the ordering of results.

Available tools

search_ifs_codebase

Full hybrid search tool that combines semantic and lexical queries to return ranked results with explanations.

search_ifs_semantic

Pure semantic search using BGE-M3 embeddings to find conceptually similar code patterns.

search_ifs_lexical

Pure lexical search using BM25S to find exact API names and keywords.

embed

Create embeddings for semantic search, building or updating the vector indexes.

analyze

Analyze the codebase to extract dependencies, API calls, and structural relationships.

calculate-pagerank

Compute PageRank scores to rank files by importance within the dependency network.

import

Import an IFS Cloud ZIP file to create a versioned catalog for analysis.