I maintain pmxt.dev, an open-source unified API for prediction markets (Polymarket, Kalshi, etc.). We essentially built "ccxt for prediction markets."

We’ve hit ~450 stars and 30k downloads. We are currently the standard in the space. Limitless.Exchange endorsed us, and the founder of ccxt has even starred the repo.

What worked: Instead of traditional marketing, we focused on "Utility Marketing." We found risk-free arbitrage opportunities between markets, wrote scripts to capture them using pmxt, and posted the open-source code on r/algotrading. Basically, "Here is free alpha -> You need this library to run it -> Download pmxt."

That strategy got us the initial wave of retail algotraders, but we've plateaued. We seem to have exhausted the "retail algo" crowd. We want to break into the next tier: becoming critical infrastructure for larger players or expanding the pie.

My Question to HN: For those who have scaled niche OSS dev tools past the "initial traction" phase: 1. Do we keep feeding the retail crowd? (e.g., more complex strategies, more "make money" scripts?) 2. Or do we pivot to "Enterprise" features? (e.g., focus purely on latency, reliability, and institutional docs?) 3. Is this just a market cap issue? Are prediction markets just too small right now to support a 1k+ star library, and we should just wait for the industry to catch up?

Thanks for the insight.

https://github.com/pmxt-dev/pmxt