Reality-check your app idea before the agent codes it

What it does idea-reality-mcp is an MCP server that acts as a pre-flight check for AI coding agents. You describe a project idea in plain English, and it queries up to six developer platforms—GitHub, Hacker News, npm, PyPI, Product Hunt, and Stack Overflow—to see how many similar tools already exist. It returns a 0–100 “reality score” with trend direction, top competitors, and pivot hints, so the agent can decide whether to build, pivot, or kill the idea before writing any code.

The interesting bit The clever part is turning market research into an automated gate inside the agent loop. By adding a single line to your agent’s instructions file, idea_check triggers automatically on new projects, effectively forcing your agent to do its homework. The scoring also adapts on the fly: if a data source is down or a token is missing, its weight is redistributed among the remaining sources so the score doesn’t collapse.

Key highlights

• Queries six sources in parallel and returns a structured reality_signal, competitor list, and AI-generated pivot suggestions.
• Two scan depths: quick (GitHub + HN, under three seconds) and deep (all six sources).
• Ships with a hosted REST API and a browser demo that need no local setup.
• Includes an onboarding CLI and a built-in health check that verify the MCP server and scoring engine.

Caveats

• The README openly warns that the tool can miss obvious competitors or return irrelevant results; accuracy hinges on keyword extraction that the authors admit is still being refined.
• Deep mode includes Product Hunt, but only if you supply an optional token; otherwise that source is silently dropped and its weight redistributed.
• GitHub API rate limits apply unless you configure an optional token, which could throttle results in practice.

Verdict Worth integrating if you use Claude, Cursor, or similar agents and want to stop accidentally rebuilding saturated tools. Skip it if you need rigorous, human-grade market analysis rather than an automated heuristic.