The deep intent layer. It reads a prospect's recent LinkedIn posts, finds the themes that map to what you sell, records a scored intent signal onto the account, and saves the quoted evidence as a note. Run it after signal scan, and only on the leads that already qualified, because this one costs money per profile.
linkedin
apify
claude
One job, four steps. Posts in, intent out. It does not scan websites and it does not write outreach.
Their last twenty posts, resolved by LinkedIn URL or email rather than by name. Around ten cents a profile, and it tells you the total before it starts.
apify
Not keyword matching. It reads the posts for whether this person is consciously working the problem you solve, and how close they are to doing something about it.
claude
A scored intent signal lands on the account, with the angle drawn from a real post. It feeds the ICP model and shows up on the signals tab.
The quotes, the post links and the themes are saved as a note, so whoever writes the first touch is quoting them rather than guessing.
Every skill is a plain markdown file in your own folder. Read it, change it, and it is yours.
The themes it hunts for come from your offer in the GTM profile. Name the problem you actually solve and the scan stops finding intent that is not intent.
Change the post count, the look back, or the qualifying threshold it will run above. Hunt one named theme across a list instead of scanning open ended.
It is the second pass on purpose. Run signal-scan first for free, then point this at the qualified leads only. For one person before a call, use meeting-brief.
Twenty posts read properly. One intent signal, and the quotes to prove it.
Paste it into your terminal. Then use /content-scan inside Claude Code.
curl -sL https://raw.githubusercontent.com/NousC/gtm-skills/main/.claude/skills/content-scan/SKILL.md --create-dirs -o ~/.claude/skills/content-scan/SKILL.md
Open Claude Code and type /content-scan. Point it at a qualified lead or a list. It shows you the profile count and the cost, then waits.