Model Context Protocol (MCP) Servers: Sequential Thinking with bonus Leave Management server.

What is Model Context Protocol (MCP) Servers: Sequential Thinking with bonus Leave Management server.?

This project implements two Model Context Protocol (MCP) servers using the FastMCP framework: Sequential Thinking for structured problem‑solving and Leave Manager for HR leave tasks. Both servers run concurrently via a single main.py launcher and are queried through Claude Desktop.

How to use Model Context Protocol (MCP) Servers: Sequential Thinking with bonus Leave Management server.?

Install dependencies (Python 3.8+, uv, mcp[cli]), place the provided main.py, leave_manager.py, and sequential.py in the project directory, then run uv run python main.py to start both servers (Leave Manager on port 8000, Sequential Thinking on port 8001). Interact via Claude Desktop by sending JSON queries to the respective endpoints.

Key features of Model Context Protocol (MCP) Servers: Sequential Thinking with bonus Leave Management server.

• Sequential Thinking: step‑by‑step reasoning with revision, branching, and dynamic step count.
• Leave Manager: tools for balance, apply, history; resource greeting://{name}.
• Concurrent launch via multiprocessing in a single script.
• Communicates via HTTP on localhost ports 8000 and 8001.
• Integrates with Claude Desktop for sending MCP queries.

Use cases of Model Context Protocol (MCP) Servers: Sequential Thinking with bonus Leave Management server.

• Plan a supply chain for a smartwatch using iterative reasoning and alternative strategies.
• Manage employee leave requests, balances, and history for HR departments.
• Break down complex projects into sequential thoughts with revision and branching.
• Automate multi‑step reasoning tasks like cost analysis or timeline planning.

FAQ from Model Context Protocol (MCP) Servers: Sequential Thinking with bonus Leave Management server.

What are the prerequisites to run these servers?

Python 3.8 or higher, Claude Desktop (for querying), and the uv package manager. The system is tested on Linux (e.g., Ubuntu).

How do I start both servers?

Run uv run python main.py from the project directory. This launches Leave Manager on http://localhost:8000 and Sequential Thinking on http://localhost:8001 simultaneously.

How do I query the Sequential Thinking server via Claude Desktop?

Send JSON payloads to http://localhost:8001/tool/sequential_thinking including fields like thought, nextThoughtNeeded, thoughtNumber, and totalThoughts. Optionally use sessionId for context, isRevision, or branchFromThought.

Is there any data persistence across sessions?

The Leave Manager uses a mock employee_leaves.json database. Sequential Thinking does not persist plans; each session uses a temporary sessionId.

What transport and authentication are used?

Both servers communicate over plain HTTP on localhost (no TLS). No authentication is enforced; they are intended for local development use.