What is mcp-discover?

It is a Claude Code skill that scans your project to suggest which parts of your code should be exposed as MCP tools or resources.

How does it ensure security during analysis?

The skill follows a strict checklist to exclude private methods, internal implementation details, and raw database access, ensuring only safe, parameterized handlers are recommended.

Does it support languages other than Node.js?

While it has strong support for Node.js frameworks like Express and Fastify, it performs general scans for database schemas, documentation, and config files applicable to various stacks.

Can it generate the actual MCP server code?

Yes, once the discovery is complete, it can hand off the results to the /mcp-init command to bootstrap a production-ready MCP server.

Why does it scan documentation first?

Scanning READMEs and API docs allows the tool to extract domain context and entity relationships, resulting in more accurate and descriptive tool definitions for the AI.

MCP Project Discovery

Name: MCP Project Discovery
Author: randyquaye

byrandyquaye

0•

API開発

Analyzes codebases to automatically identify and recommend tools, resources, and prompts for Model Context Protocol (MCP) servers.

MCP Project Discovery is a specialized skill designed to streamline the creation of enterprise-grade Model Context Protocol (MCP) servers by performing a deep audit of existing codebases. It automatically scans project structures, documentation, API routes, and database models to extract domain context and suggest optimal tool definitions. By classifying findings into tools, resources, and prompts while applying security filters to exclude sensitive data, it provides a comprehensive blueprint for extending Claude's capabilities with specific project logic.

主な機能

01Automated codebase auditing for MCP capability identification

02Direct integration with /mcp-init for seamless server generation

03Automatic classification of functions into Tools, Resources, and Prompts

040 GitHub stars

05Security-aware filtering to exclude private methods and raw DB access

06Documentation-first analysis to extract domain terminology and workflows

ユースケース

01Converting an existing Express or Fastify API into a standardized MCP server

02Quickly bootstrapping an MCP server schema based on Prisma or Drizzle database models

03Auditing a legacy codebase to determine which business logic should be exposed to AI agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add randyquaye/forge-mcp mcp-discover

For use in Claude.ai and ChatGPT