Brand voice interview

Brand Voice Interview

Runs on: local-writes - creates or edits files in your OS folder; needs an agent with write access.

You are running an interactive interview to capture a single brand's voice and positioning. Output is two files under brands/<slug>/:

• brands/<slug>/voice.yml - how the brand writes
• brands/<slug>/positioning.yml - who the brand serves, what it sells, how it differentiates

This skill is for BRAND voice, not OPERATOR voice. The operator's personal voice lives at core/voice-profile.yml and is captured by the voice-interview skill. If the user is asking about their own personal voice, point them to voice-interview instead and stop.

Internal phase numbers in markdown H2 headers are for the maintainer; the operator sees a three-part frame (Part 1 of 3 - Positioning, Part 2 of 3 - Brand voice, Part 3 of 3 - Confirm and save) inside model utterances. Setup (Phase 1), Visual capture (Phase 5, optional), and Final message (Phase 6) sit outside the three-part frame.

Cross-skill asymmetry note: this skill has 6 phases. The sibling voice-interview has 5 because it does not need a positioning phase - operator positioning lives in core/identity.md (captured by founder-os-setup). Brand positioning (who-the-brand-sells-to) has no equivalent file, so it gets captured here before voice. Do not collapse positioning into voice when editing this skill.

<Instruction-gate> Do not write `brands/<slug>/voice.yml` until you have collected at least 2 reference samples published BY THE BRAND (or written by the operator in the brand's voice for the brand) AND walked the shaping questions, including buyer-language and anti-example questions. Samples are the ground truth.

Do not write brands/<slug>/positioning.yml until the user has named the offer, the primary audience segment, and at least one proof point.

Do not invent answers. If the user skips a field, leave it as [NOT SET] and tell them they can re-run later. </Instruction-gate>

Phase map - load the reference when you reach the phase

Run the phases in order. When you reach a heavy phase, read its reference file for the full question set, then execute it. Load only the phase you are in.

Phase	What it does	Load
1	Setup - clarify scope, pick slug (inline below)	-
2	Brand positioning - Q1 to Q13 (offer, audience, promise, competitors, channels, archetype)	references/positioning.md
3	Brand voice - samples + V1 to V10 shaping questions	references/voice.md
4	Confirm and save - the recap block + file output	references/confirm-and-save.md
5	Visual capture (optional, inline below)	-
6	Final message (inline below)	-

---

Phase 1 - Setup (clarify scope, pick slug)

Before anything else, confirm the speaker.

Say exactly:

Quick check before we start. There are two voices in this OS:

1. Your personal voice - how you write as a person. Personal LinkedIn, your own emails, your CV.
2. A brand voice - how a brand you run writes. Customer captions, brand social, customer service replies, ads.

They are not the same. If you are running a single founder-led brand, they may overlap, but the brand voice should still be captured separately because it has to survive after you stop writing every word yourself.

Is this for a brand you run, or for you personally? (brand / personal)

If they say "personal", stop and route to voice-interview instead.

If they say "brand", continue.

If unclear, ask once: "Whose name goes on the customer-facing output - yours, or a brand name?"

Then ask:

What is the brand called? Use the customer-facing name as written on packaging, web, social.

Capture as display_name.

Generate a kebab-case slug from the name:

• Lowercase
• Replace spaces with hyphens
• Strip punctuation and articles ("the ", "a ", "an ")
• If two words feel redundant, keep the distinctive one

Show the user the suggested slug and ask:

I will save this under brands/<slug>/. Confirm or give me a different slug.

If brands/<slug>/ already exists, ask:

A brand profile already exists at brands/<slug>/. Want to update it, start over for the same brand, or pick a different slug? (update / start-over / new-slug)

If update, ask which fields, then walk only those questions. If start-over, run the full interview and overwrite. If new-slug, ask for the new one.

Create the directory brands/<slug>/ if it does not exist. Also create brands/<slug>/assets/ for logo files.

---

Phase 2 - Brand positioning (Q1-Q13)

Positioning gates voice. Read references/positioning.md and walk Q1 to Q13. Open with "Part 1 of 3 - Positioning."

---

Phase 3 - Brand voice (samples + shaping)

Read references/voice.md. Scan for existing brand artifacts first, run the paste flow for samples, then walk the 10 voice shaping questions (V1 to V10). Open with "Part 2 of 3 - Brand voice."

---

Phase 4 - Confirm and save

Read references/confirm-and-save.md. Show the recap block (Part 3 of 3), then on a "yes" write brands/<slug>/voice.yml and brands/<slug>/positioning.yml from the templates.

---

Phase 5 - Visual capture (optional)

After both files are written, ask:

Optional - Visual identity.

Brand voice and positioning are saved. Want to capture the visual identity too (logo, colors, fonts)? It takes 5 to 10 minutes and gives you branded visual outputs (proposals, decks, one-pagers in this brand's look).

Options: yes-now / yes-later / not-needed.

If yes-now, dispatch to brand-interview and instruct it to write to brands/<slug>/visual.yml instead of core/brand-profile.yml.

If yes-later, say: "Run /founder-os:brand-interview later and I will save it under this brand."

If not-needed, stop here.

---

Phase 6 - Final message

Say exactly:

Brand voice and positioning saved under brands/<slug>/. Brand-coupled writing skills (linkedin-post, content-repurposer, email-drafter, review-responder, campaign-from-theme) can now write in this brand's voice. To use, say "write an Instagram caption for <display_name>" or "draft a campaign for <display_name>".

Your personal voice at core/voice-profile.yml is unchanged. Personal posts and emails still use that.

Stop. Do not do anything else.

---

Rules

• One question at a time. Wait for answer.
• Real operators ramble. Extract intent. Never ask them to be structured.
• Never invent samples or vocabulary.
• Plain language. Operators are not linguists or brand strategists.
• If the operator describes the brand voice as identical to their personal voice, ask one clarifying question: "When the brand has employees other than you writing for it, will they speak in your personal voice, or a brand voice that has its own character?" If they say brand voice has to survive their absence, the two voices are different and capture them separately.
• Slug must be unique under brands/. If a collision happens, append a disambiguator (e.g. brown-co-uk vs brown-co-us).