Chat mode imported from JoshuaWink/vscode-api-expose-extension (.github/chatmodes/demo.chatmode.md). Copyright stays with the author.

DeepCog Disciplined Foresight: Modularizing capi-mcp-server with fastmcp-template Patterns

Context

• Project Goal: Expose the VSCode capi CLI as a Model Context Protocol (MCP) server, enabling agent-driven automation and observability with minimal human intervention.
• Current State:
  • capi-mcp-server is a single-file Python MCP server using FastMCP and direct tool decorators.
  • The fastmcp-template provides a modular, registry-driven, and extensible server architecture.
• Intent:
  • Refactor capi-mcp-server to use the modular, registry-based approach from fastmcp-template for long-term maintainability, extensibility, and agent compatibility.

Why Disciplined Foresight?

• Reason: Ensures the refactor is logically sound, modular, and robust for future growth.
• Memory: Leverages best practices from the template and prior MCP/LLM server architectures.
• Will: Drives purposeful action—scaffolding a maintainable, extensible, and agent-friendly server.

Step-by-Step Plan

1. Directory Structure:
   • Create app/ with tools.py, resources.py, prompts.py, and __init__.py.
   • Move tool logic from server.py to app/tools.py.
   • Add a register_tools(server) function for dynamic tool registration.
2. Entrypoint Refactor:
   • Update server.py to instantiate FastMCP, call register_tools, and run the server.
   • Prepare for future register_resources and register_prompts.
3. Registry Pattern:
   • Use a list of tool definitions (name, description, factory) for easy extension and testability.
   • Register each tool dynamically in register_tools.
4. Extensibility:
   • The same pattern can be used for resources and prompts, supporting future agent workflows and LLM integrations.
5. Testing & TDD:
   • Structure supports adding a tests/ directory for incremental, test-driven development.

Recap TL;DR

• Refactor capi-mcp-server to use a modular, registry-driven architecture inspired by fastmcp-template.
• Goal: Long-term maintainability, extensibility, and agent/LLM compatibility.
• Pattern: Move tool logic to app/tools.py, use register_tools, and keep server.py minimal and declarative.
• Outcome: A future-proof, agent-friendly MCP server ready for rapid iteration and onboarding.

This document was generated by DeepCog using the Disciplined Foresight perspective to ensure strategic, maintainable, and extensible architecture for agent-driven automation.