delinea-mcp AGENTS.md

DelineaMCP Development Guide

This document provides AI coding agents with project-specific guidance for the DelineaMCP (Model Context Protocol) server that integrates with Delinea Secret Server and Platform APIs.

Project Overview

DelineaMCP is an MCP server built on FastMCP that provides AI assistants with secure access to Delinea Secret Server and Platform APIs. The system supports both STDIO transport (CLI tools like Claude Desktop) and SSE transport (web-based integrations like ChatGPT) with optional OAuth 2.0 authentication.

Development Commands

Package Manager: Use uv as the primary interface for all Python operations. This project heavily relies on uv for dependency management, virtual environment setup, and command execution.

Setup and Installation

# Install dependencies (preferred)
uv sync --frozen

# Alternative dependency installation
uv pip sync requirements.txt

Running the Server

# Development mode (STDIO transport)
uv run server.py

# With custom configuration
uv run server.py --config config.json

# SSE mode with OAuth (requires config)
uv run server.py --config production-config.json

Testing

# Run unit tests (excludes integration tests)
uv run pytest -q --ignore=tests/integration/ tests/ -k "not test_live"

# Run with coverage
uv run coverage run -m pytest -q --ignore=tests/integration/ tests/ -k "not test_live"
uv run coverage report --omit="tests/*" --show-missing

# Run specific test file
uv run pytest tests/test_tools.py

# Integration tests (requires credentials)
export DELINEA_PASSWORD=<password>
export LIVE_SECRET_ID=<id>
uv run pytest tests/integration/

Build Commands

# Compile dependencies (when updating pyproject.toml)
uv pip compile pyproject.toml --universal -o requirements.txt

# Lint check
uv run python -m py_compile server.py
uv run python -m py_compile delinea_api.py

Linting && Formatting

Prior to any completion of work, ensure code linting and formatting is applied with the trunk tool.