luminarforge.
Try it free
Writing

Written in their voice.
Scored before it posts.

Their website teaches it the brand. Three writing samples set a measured voice. A writing team drafts, edits and scores every post; you approve it.

Try it free See all the examples Free to start. 5,000 credits, about eight posts. No card.
A voice profile measured from three of the brand’s own pages
A voice, measured from three of their own pages. Example workspace shown: Luminar Works.
The thing working

A scored draft, and what the score is for.

The post is scored on eight dimensions — how specific it is, how it reads, AI tells, the hook, platform fit, expertise. The score is not a gate. It shows you where the draft is weak; you decide what publishes.

A finished run: the draft on the right, the console below with every bench done
One run, as it finished: the draft on the right; below, Strategist, Writer, Editor and Humaniser each reporting what they did.
A generated LinkedIn draft with its quality score and the dimensions beneath it
A real draft and its score, with the weak dimension named. Nothing is blocked by it.
The writing team’s console mid-run: Strategist done, Writer done, Editor reviewing
The console during a run: Strategist done, the Writer’s draft in, the Editor reviewing for depth, specificity and flow.

Example workspace shown: Luminar Works.

The writing team

Five agents, not one prompt.

Every draft goes through all five; there is no faster setting to forget to switch off. You watch the first four work.

  1. 01

    Strategist

    picks the angle from the brand and the brief

  2. 02

    Writer

    drafts in the measured voice

  3. 03

    Editor

    tightens

  4. 04

    Humaniser

    strikes common AI phrasing

  5. 05

    Condenser

    trims to the platform’s length

Writing a post: about 146 credits, before its picture.

What it does

The brand from a URL. The voice from three samples.

  • The knowledge base, read from a website into sections

    It learns the brand from the website

    Paste a client’s website. The wizard reads it into pages, and the writing team reads those pages on every post.

  • Voice characteristics measured from the samples: formality, humour, storytelling, metaphor

    A measured voice from three samples

    Paste three pieces of their real writing. Ten style metrics are measured without an AI call, and one AI pass names traits like formality and humour. The Writer drafts inside them.

  • One scoring dimension opened: burstiness, cliché density, formatting tells

    Fewer AI tells

    Every post is scanned at scoring against a library of phrases that read as machine-written, and the rewrite pass works from part of that list.

Example workspace shown: Luminar Works.

Step by step

From a website to an approved post.

  1. 01

    Paste the site

    The knowledge base builds itself. Edit anything it got wrong.

  2. 02

    Give it three samples

    Three pieces of their real writing, under Tone DNA. The voice is measured from those.

  3. 03

    Brief the post

    A topic and a platform. The team drafts, edits and scores it.

  4. 04

    Approve

    Edit the text, take the score’s notes or not, and send it to the calendar.

Questions

What people ask.

Do I need a voice profile to start?

No. Without one the team writes from the knowledge base alone. Add three samples under Tone DNA when you have them.

Is the score a pass or a fail?

Neither. Nothing is blocked by it. It shows where the draft is weak, dimension by dimension; you decide what publishes.

What does the writing cost?

About 146 credits for a typical post, before its picture. Regenerating is charged again.

Can I edit the draft?

Yes, at any point. The team writes the first version; the post is yours.

Does it post by itself?

No. Nothing publishes until you approve it.

Next

Now put a real picture beside it.

Try it free See real people & products See templates