I open-sourced my LinkedIn prospect research tool as a Claude Code plugin
Heat trend
The percentage is based on available heat signal, not comment count or independent people.
A developer open-sourced a LinkedIn prospect research tool as a Claude Code plugin, available under an MIT license. This tool scrapes LinkedIn posts, profiles, and comment threads into a local SQLite database using Apify. It classifies content, identifies individuals discussing relevant problems in comment threads, and discovers shared uncommon commonalities from profiles. The tool also logs outreach efforts and learns which hooks generate replies, emphasizing local data storage and no telemetry.
I do outreach for my startup. Started with Apollo and Clay doing mass automated messaging — conversion was bad. Switched to fewer, manually written, better targeted messages and reply rates went up a lot.
The bottleneck then became research: reading dozens of posts to work out what someone actually cares about. So I automated that part. Not the sending.
What it does - Scrapes LinkedIn posts, profiles and comment threads into local SQLite (via Apify) - Classifies what a person or company actually posts about - Mines a post's comment thread for people already describing your problem - Finds uncommon commonalities from full profiles — shared employers, schools, volunteer work - Logs what you sent and what came back, then learns which hooks get replies
18 MCP tools + 8 skills.
/plugin marketplace add spirosbax/insaight
/plugin install insaight@insaight
MIT, no telemetry, everything stays local: https://github.com/spirosbax/insaight