Hey HN!

I built Pi Exa initially make use Exa MCP's free 1000 reqs/day for web search + fetch, but since then added Exa's deep search capabilities since it was pretty useful for finding and reading research papers!

Basic search and fetching work without an API key, and the extension connects lazily on the first tool call and caches the MCP schemas, so Pi can register the tools at startup without establishing a connection.

For those who want granular control over the extension and your context window, you could also turn off certain tools and even the whole extension!

Would appreciate feedback on which search/research workflows would be most useful for people!