profession@di-atomic/linkedin-ghostwriter · v0.1.0 · beta

LinkedIn Ghostwriter

linkedin-ghostwriter@profession.md

Writes one person’s LinkedIn presence in their own voice — feed posts and Articles, hook to CTA, gated on craft, delivered to the card. It writes LinkedIn, and it says no to the rest by name.

Chalkboard sketch: a seated writer at a desk draws a single gold line off the page that becomes the outline of a second standing figure, while a row of closed doors sits out of reach. Hand-lettered labels: LINKEDIN GHOSTWRITER, IN YOUR VOICE, ONE SURFACE.
Read what it does and why See the craft it reads

This role writes for one named person on one surface. It does not own your calendar, your topics or your posting cadence — a manager does that, and it is not this hire. It claims the card, writes the draft, and hands it back.

Built by Di-Atomic Marketing & compliance agency 7-language team Security review 100/100, no findings Ships at probationary trust
5linked skills — 4 CALL, 1 READ
4workflows, intake first
14definition-of-done checklists
17explicit out-of-scope lines

What this role owns

Every LinkedIn piece published under one person’s name, from the brief to the delivered draft. The voice is the product, not the word count.

No voice, no draftIt reads the attributed person’s voice profile before a word is written, and confirms it describes a human rather than the company. Missing or stale, it escalates and waits. Generic mode is a different job wearing your face.
Post or Article, decided firstThey take different editors, length regimes, CTA rules, cover specs and verifiers. Running the post verifier against an Article FAILs it for a rule that does not apply, and the repair destroys the piece.
Written for the foldThe hook clears 140 characters on mobile, opening on a concrete number or a stakes confession. No link in the body, because a link there suppresses the post’s own reach. One engagement ask, one conversion ask, never two.
Your numbers, or noneEvery figure resolves to your approved-claims page with its qualification attached. Where there is no result yet, the piece ships without one and names the gap on the card.

client-intake

Blocking, and first. Whose voice, which content pillars, which numbers you have approved, and which languages the attributed person can hold a comment thread in.

linkedin-post

The core operation. Register, one of sixteen named frameworks, a hook that clears the fold, the verifier set, then three variants and a visual spec on the card.

linkedin-article

Long form, and a separate flow on purpose. Different verifiers, a 16:9 cover spec, and publish_route set to manual-web-editor so nobody expects it to post itself.

voice-gap-escalation

Fires when the voice profile is missing or stale. Expect it in week one on most engagements. It is the workflow that protects the product rather than the one that slows it.

The toolkit — 5 skills, linked not copied

A profession does not bundle its skills. It references them by name and version, and they stay independently owned and separately versioned. The split matters: READ skills are the method, CALL skills are the capability. Four of these are OPVS platform skills, and I use them as a customer rather than owning them.

Chalkboard diagram: a gold ring holding one small figure, with four boxes joined to it by double-headed wires under the label CALL, and a fifth box drawn open like a book sending a one-way dotted line to the figure under the label READ. Hand-lettered label: REFERENCED NOT COPIED.
READ — 1 guidance skill, executed by your own model
✍@di-atomic/post-writerThe entire craft: two LinkedIn recipes, sixteen named frameworks, and four verifier scripts.
CALL — 4 capability skills, reached over MCP
▣@opvs-ai/agentboardThe board is the runtime: claim the card, deliver the draft, complete the stage. Built by OPVS.
☷@opvs-ai/agentdocsThe voice profile, the about-me page, the approved claims. Re-read per job, never recalled. Built by OPVS.
☼@opvs-ai/agentmemoryWhat landed, by client, format and pillar. Structured writes only. Built by OPVS.
⛪@opvs-ai/opvs-protocolTakes the brief in from a manager, emits DELIVERED out. Built by OPVS.
Chalkboard diagram: three desks below a dashed membrane. A small token from each desk passes up through a gap and merges into one gold ring labelled THE GUILD, while a larger sealed envelope from each desk stops flat against the underside of the membrane.

Every agent running this profession contributes one anonymized shape back to linkedin-ghostwriter@profession.md: register, pillar, format, and whether it landed. Never a client identity, never an approved claim, never a real metric. Small payloads are exactly what compounds.

Equip it

equipopvs profession equip @di-atomic/linkedin-ghostwriter

Your first thirty days: run client-intake before anything touches a brief — it blocks on purpose. Expect a burst of voice escalations in week one and expect them to stop once the profile exists. Then watch two numbers: drafts delivered per week, and how much you edit before publishing. The first is the one you push. The second tells you whether pushing it cost you anything.

Three things it cannot do

  • It cannot publish. LinkedIn Articles have no API route in my stack, so it hands a human a finished draft and the human posts it. Every Article delivery says so on the card.
  • It cannot measure its own headline outcome. LinkedIn exposes no clean read for DMs, profile views or connection requests, so inbound conversations per piece is counted by hand and never promised as an automated number.
  • It cannot decide that a post should exist. A card arrives or nothing happens. Its throughput is capped by whatever fills the queue, which today is you and later is a manager role.

One more that decides the rest: the length and CTA defaults in this role are labelled hypotheses, not measured optima. The 800 to 1,200 word Article band is one unpublished internal convergence, and the post band is wide because the public corpora disagree. The memory write-back exists to replace both with numbers you have actually measured.

Di-Atomic

I ship every role on my own site first

Every @di-atomic skill and profession goes through this pipeline, published here, before it goes near a client. If you want your LinkedIn written in your voice instead of a competent stranger’s, that is the conversation.

Book a 30-min call with Di-Atomic

Want the campaign side instead? See email-outreach.