N’012 — MCP Find — Open-Source Server Directory.
MCP Find — Open-Source Server Directory
Design and frontend for the first open-source, AI-agent-optimized MCP server directory — built by two people in one week with zero budget, competing against 6+ closed-source incumbents.
Client
Project Type
Timeline
Role
Overview
Launched as the first open-source MCP directory with a real UI, MIT-licensed. Designed two compounding distribution flywheels: a fast flywheel (launch narrative driving HN/Reddit traction, GitHub stars, and social proof) and a slow flywheel (llms.txt + JSON-LD + SSR enabling AI agent discoverability). Coordinated multi-platform launch across Hacker News, Twitter/X, Reddit, Discord, and LinkedIn. First-mover on the AI-agent-optimized niche that no closed-source competitor had addressed.
The Brief
MCP (Model Context Protocol) is the open standard for connecting AI models to external tools — adopted by Anthropic, OpenAI, Google, and Microsoft with 97M+ monthly SDK downloads. The ecosystem has thousands of servers, but discovering, evaluating, and installing them is painful. Six-plus directories existed, all closed source, none optimized for AI agent discoverability, and none offering llms.txt or structured data for machine consumption.
Two-person team, one-week build, zero budget. The architecture had to be SSR from day one — client-rendered pages would lose both SEO and AI discoverability before launch. Distribution infrastructure (SEO metadata, llms.txt, JSON-LD, sitemap) had to be built into the product during development, not bolted on after. The directory needed to differentiate on four axes simultaneously: open source, AI-agent optimization, config snippet generation, and tool schema visualization.
Designed and built both the product and its distribution infrastructure simultaneously — the architecture decisions were inseparable from the go-to-market strategy.
Decisions
Tradeoffs
Building distribution infrastructure (SEO metadata, llms.txt, JSON-LD, sitemap) during a one-week sprint meant less time for polish on secondary pages. Accepted that tradeoff because discoverability was existential — a polished product no one can find loses to an adequate product that compounds.
Next Project
Hospital Catalog Equivalency Search