PeopleSoft MCP Server

A minimal Model Context Protocol (MCP) server, built as a worked example of how an MCP server is constructed. It exposes a local SQLite contracts/billing database (PROJECT, CA_DETAIL, CA_BILL_PLAN) through a small set of tools, ranging from generic schema introspection to purpose-built semantic queries.

Features

• Schema introspection tools — discover tables and columns without any prior knowledge of the schema

• Semantic tools — purpose-built queries for the contracts/billing domain (projects, contract lines, bill plans)

• Direct SQL tool — an escape hatch for arbitrary queries once the schema is known

• Direct database access via Python's stdlib sqlite3, wrapped in asyncio.to_thread for async compatibility

Related MCP server: MySQL MCP Server

Quick Start

Prerequisites

• Python 3.11+

• uv package manager (recommended)

Installation

# Clone the repository
git clone <repo-url>
cd peoplesoft-mcp

# Install dependencies
uv sync

# Create finstg.db and load static data
uv run create_db.py

Configuration

1. Copy the example environment file:

cp .env.example .env

2. Edit .env if you want to point at a different SQLite file (defaults to finstg.db in the repo root):

SQLITE_DB_PATH=finstg.db

3. Copy .cursor/mcp.json.example to .cursor/mcp.json and update the path to your installation.

Running the Server

uv run peoplesoft_server.py

Cursor IDE Integration

The MCP config (.cursor/mcp.json, copied from .cursor/mcp.json.example) should look like:

{
  "mcpServers": {
    "peoplesoft": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mcp_ps/",
        "run",
        "peoplesoft_server.py"
      ]
    }
  }
}

Available Tools

Schema Introspection (2 tools)

Contracts & Billing Module (4 tools)

Billing Analysis (2 tools)

Direct Query (1 tool)

Project Structure

peoplesoft-mcp/
├── peoplesoft_server.py    # Main MCP server entry point
├── db.py                   # SQLite connection management
├── create_db.py            # Creates finstg.db and loads data/*.csv
├── db_structure.txt        # Schema definition source for create_db.py
├── data/                   # Static CSV data loaded into finstg.db
│   ├── project.csv
│   ├── ca_detail.csv
│   ├── ca_bill_plan.csv
│   └── proj_resource.csv
├── finstg.db               # Local SQLite database (generated)
├── tools/                  # Semantic tool modules
│   ├── introspection.py    # Schema discovery tools
│   ├── contracts.py        # Contracts/billing tools
│   └── billing.py          # Billing setup & cost-reimbursable analysis tools
├── tests/                  # Test suite
│   └── test_contracts.py
└── pyproject.toml          # Project configuration

Running Tests

uv run pytest tests/ -v -s

Example Queries

• "List all active projects for business unit UCD"

• "What contract lines and bill plans belong to project 25B1111?"

• "What's the bill plan status for contract FAKE4 line 2?"

• "What tables and columns are in finstg.db?"

Development

Adding New Tools

1. Create a new module in tools/ or add to an existing module

2. Define async functions that use db.execute_query()

3. Add a register_tools(mcp) function

4. Import and register in peoplesoft_server.py

License

MIT

Changelog

v0.3.0 (2026-07-15)

• Removed the legacy PeopleSoft HCM Oracle tool modules (hr.py, payroll.py, benefits.py, performance.py, peopletools.py), their docs, their gated/skipped tests, and the Cursor agent/skill configs built around them

• Removed the 4 MCP resources that served the now-deleted Oracle-schema documentation

• Repurposed the repo as a standalone worked example of MCP server construction (schema introspection + semantic tools + direct-SQL escape hatch) over a local SQLite database

v0.2.x (2026-03-02 – earlier)

• Replaced Oracle (oracledb) backend with a local SQLite database (finstg.db)

• Rewrote schema introspection to use SQLite's own metadata (sqlite_master, PRAGMA table_info)

• Added tools/contracts.py with semantic tools for the local PROJECT/CA_DETAIL/CA_BILL_PLAN schema