mnemox-ai/idea-reality-mcp

What it solves

This project provides a way to automatically check if a software or startup idea already exists before writing any code. It prevents developers from building redundant tools by providing a competitive analysis and a "reality score" to determine if a market is too crowded or growing.

How it works

It operates as a Model Context Protocol (MCP) server that AI agents can call. When given a description of an idea in plain English, the tool scans five different databases in parallel: GitHub repositories, Hacker News discussions, npm packages, PyPI packages, and Stack Overflow questions. It then calculates a reality score (0–100) based on weighted evidence from these sources, detects trends (accelerating, stable, or declining) and provides AI-generated pivot suggestions.

Who it’s for

Developers, entrepreneurs, and AI agent users who want to integrate market research into their development workflow to decide whether to build, pivot, or kill a project idea.

Highlights

• Multi-source scanning: Checks GitHub, HN, npm, PyPI, and Stack Overflow simultaneously.
• Two search modes: Offers a "quick" mode for fast sanity checks and a "deep" mode for comprehensive competitive scans.
• Agent-native: Designed to run inside AI agents (like Claude or Cursor) so the research happens automatically.
• Comprehensive output: Returns a reality signal, market momentum, top competitors, and pivot hints.
• Easy installation: Supports various MCP clients including Claude Desktop, Cursor, Windsurf, and Cline.