Installs into .claude/skills of the current project.
Are you the author of Linkedin?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/tomcounsell-linkedin-ai)
---
name: linkedin
description: "Read LinkedIn posts or DMs, draft content, and perform specifically requested posting or engagement actions."
---
# Linkedin
Use the available authenticated browser surface or LinkedIn connector. Confirm the signed-in account and reuse the relevant tab. Read the complete post or DM thread before deciding what to say; the feed's visible accessibility controls may omit post-body text, so inspect the post itself. Refresh state after each UI change and verify the actual submit result.
Follow the requested scope. A broad request to check LinkedIn means inspect and summarize; it does not authorize new posts, DM replies, likes, or follows. If publishing or replying is explicitly authorized, prepare the content and continue without asking for the same authorization again.
For posts based on recent work, inspect real commits or supplied evidence and extract a portable lesson. Remove private client details and internal identifiers unless intended for that audience. Use plain, grounded prose with a point of view; avoid generic congratulations and engagement quotas. Run $de-slop then $authenticity-pass before publication. A failed gate gets at most two revision cycles.
Use observed UI locators, not copied BYOB indices or old CSS selectors. If the composer is unavailable, retain the draft and report the failure. Verify the published text and permalink or message state before claiming success; uncertain submissions require a read before retrying.
