LinkedIn Ghostwriter

linkedin-ghostwriter is the hire a founder makes when they know LinkedIn is where their buyers are, have decided they will not write there themselves, and do not want an agency retainer. Today that role costs $2-5k a month as a human. What it buys that installing the underlying skill does not is judgment: @di-atomic/post-writer is 237 files across eight platforms, and installed raw it is a library, leaving the buyer to know which of sixteen LinkedIn frameworks fits a career-arc story, that the hook has to clear a 140-character fold, that a comment-gate is banned on an ICP-pillar Article, and which of four verifier scripts grades an Article rather than a post — getting that last one wrong FAILs an 850-word piece for violating a 3,000-character ceiling it was never subject to. The profession is the layer that already knows all of that, and its refusals are what make it reliable: it will not quietly wander into an Instagram caption because the same skill happens to contain one, it will not invent a metric the client never approved, and it will not write in a generic voice when the voice profile is missing — it escalates, names the artifact needed, and waits. It is deliberately not the thing that builds that voice profile either. Linking voice-builder would make it construct the register and then grade its own conformance to it, which is the same failure as an agent writing its own guard. The honest limits ship with it: it cannot publish, because LinkedIn Articles have no API route in our stack and a human pastes them; it cannot measure its own headline outcome, because LinkedIn exposes no clean read for DMs or connection requests, so that number is counted by hand and must never be promised as an automated one; and it cannot decide that a post should exist at all. A card arrives or nothing happens. It writes one surface, for one person, extremely well.