How I built an MCP directory with AI agents

I went from 0 to ~100,000 pageviews a month in 30 days thanks to my MCP directory

(here's how I built it with AI agents)

First, the top line stats for the humblebrag:

- 35,000 visitors
- 92,000 page views
- 1,500 email subscribers
- 18,000 visitors from Google
- 10,000 direct visits
- 4,378 visitors from X
- 262 visitors from Perplexity
- 169 visitors from ChatGPT

MCP directories are being launched all of the place, and many honestly probably better looking that mine

but the secret weapon I have is programmatic SEO and knowing how to take the data and turn it into content that ranks

the process looks like this:

1. a custom scraper built by Gemini 2.5 crawls open source MCP servers to find any that don't exist in my app and creates them

2. the scraper passes it off to an agent (Claude 3.7) that is trained to write content that ranks, specifically focused on how to set up the server and use the tools it provides

3. while that happens another bot is fetching stats from different services like npm download stats

4. every day, the sources are checked to update the stats and if the details of how the server works changes, the article is rewritten to cover it

This entire thing was built on a saturday morning and hasn't really been touched since

Next up, I plan to use AI to categorise these and build more landing pages for different types of server, but so far, so good

Anybody wanna advertise on the directory? ;)