brand-context: research and fill your brand files

Setup Cowork-ready

What it does

brand-context is your onboarding engine. Give it a URL, and it researches your brand from the outside in: who your buyer is, what language they use, what your competitors are doing, how to position your offer, and what psychological angles move people to act. Then it writes all of that into structured files that every other skill reads from.

Without those files, the toolkit produces generic output. With them, it produces work that sounds like someone who has studied your brand for months. Run brand-context once when you set up a new brand, and all the downstream skills get sharper automatically.

It covers 11 intelligence files in a single session: brand guide, buyer personas, psychographic angles, hooks by awareness level, value props, objection map, competitive intel, offer deep dive, voice DNA, case studies, and testimonials.

When to reach for it

You’re onboarding a new brand and want every skill pre-loaded with real intelligence.
You took on a new client and need to get up to speed fast.
Your brand files are stale or empty and your copy has felt generic lately.
You want to audit a competitor: their positioning, angles, and gaps.
A skill produced output that felt off, and you suspect the underlying intelligence files need a refresh.

Where it runs

Cowork ✓ Desktop ✓

brand-context works in both environments, with one nuance worth knowing: the part that pulls your exact colors, fonts, and logo directly from your live website uses a browser. That step is most reliable on the desktop app. Everything else (market research, personas, psychographic angles, competitive intel) works equally well in Cowork or the desktop app.

If you’re in Cowork and the visual extraction step can’t run, it will pause and ask you to paste your brand colors and fonts instead of guessing.

How to run it

The easiest path is the friendly walkthrough in Onboard your brand. That guide collects everything up front so brand-context can run with fewer back-and-forth pauses.

You can also trigger it directly:

“Set up brand context for examplebrand.com”

Or be more specific about what you’re doing:

“Onboard a new brand: examplebrand.com: I manage multiple properties.”

If you’re continuing a setup that was interrupted:

“Resume brand-context setup for examplebrand.com: we stopped partway through.”

It checks which files are already populated and picks up from the first gap rather than starting over.

What it needs from you

A brand URL. The more live and complete the site, the better the research.
(Optional but valuable) 5 to 10 writing samples in the brand’s voice. These are the primary source for the voice DNA file. If the brand has a distinct tone, pasting a few real examples produces much sharper results than research alone.
Your real case study metrics and approved customer quotes, if you have them. brand-context never fabricates proof. It will build the placeholder files and ask you to fill them in rather than inventing numbers.
For offer details not visible on the site (price, guarantee, qualification criteria): it will ask you as it hits those sections.

What you get back

Eleven populated intelligence files saved to your brand folder:

brand-guide.md: colors, fonts, logo, visual mood, positioning, social proof signals.
buyer-personas.md: five micro-personas with demographics, psychographics, decision triggers, and the objections each one carries.
psychographic-map.md: 25 to 35 creative angles across seven psychological layers (financial pain, emotional pain, limiting beliefs, trigger moments, dream states, identity desires, private moments).
hooks-by-awareness.md: 50 or more hooks mapped to each Schwartz awareness level, tagged by format and persona.
value-props.md: 18 tiered value propositions (core differentiators, supporting advantages, emotional benefits).
objection-map.md: the top 10 stated objections plus three silent fears, each with a reframe and a proof point.
competitors.md: the top five competitors mapped by positioning, strengths, weaknesses, and the gap your brand can own.
offer-deep-dive.md: the full offer structure, pricing psychology, guarantees, and sales process.
voice-dna.md: sentence patterns, vocabulary, tone, structural habits, and a banned-phrases list for that brand.
case-studies.md: your verified results (filled from your input, never fabricated).
testimonials.md: your approved customer quotes (filled from your input, verbatim).

Every finding carries a confidence label: [VERIFIED] (from the brand’s own site or materials), [INFERRED] (derived from market signals), or [ESTIMATED] (best guess, flagged to confirm). Research gaps are flagged as [RESEARCH GAP] rather than filled with invented data.

After each file, it pauses, summarizes what it found, and asks you to confirm before moving on. If something is wrong, you catch it early rather than at the end of a full build.

A worked example

You: “Set up brand context for examplebrand.com.”

brand-context: (after extracting visuals and running deep research)

“Here’s what I found for the brand guide. Primary color is #1A3A5C (navy), accent is #F4A623 (amber). Font stack: Freight Display for headlines, Freight Text for body. The site positions the brand as a ‘boutique firm for founders who’ve been burned by agencies before.’ [VERIFIED]

For buyer personas, I built five micro-profiles. The dominant one is an e-commerce founder, 35 to 50, spending $30K or more per month, who had a bad agency experience and now manages ads in-house but is stretched thin. [VERIFIED, from site copy + review mining]

One research gap: I couldn’t verify pricing from the public site. Can you share the offer price and any guarantees so I can fill in offer-deep-dive.md accurately?”

That early flag means the offer file gets filled with real numbers rather than a placeholder you discover six weeks later when a page goes live.

Tips & gotchas

Let it ask before you volunteer everything. brand-context collects what it needs file by file. You don’t have to prep a full brief upfront. The Onboard your brand walkthrough shows the smoothest order if you want to prep ahead.
Voice samples matter more than you’d expect. A brand with a strong, distinctive voice (punchy and profane vs. warm and clinical vs. academic and precise) needs real writing examples, not just site copy. Pasting five to ten real samples from emails, social posts, or long-form content produces a voice-dna.md that’s actually usable. Site copy alone often sounds like every other brand in the category.
Don’t skip the review step. After each file, it shows you what it found. Read it. A wrong persona or a misidentified core offer will show up in every piece of copy and creative downstream until you fix it at the source.
Partial setups are fine. If the session gets interrupted, all written files are saved. When you come back, tell it to resume and it picks up where it stopped.
Case studies and testimonials are always yours to fill. brand-context never invents metrics or quotes. Those two files stay as [NEEDS INPUT] until you supply real data.
Colors need a browser on the desktop app for best results. In Cowork, have your brand hex codes handy as a backup. If it can’t extract from the live DOM, it will stop and ask rather than guessing.

brand-context is the engine behind onboarding. For the guided setup experience, start with Onboard your brand. Once your files are populated, apollo can go deeper on any individual module (competitive intel, voice-of-customer mining, a specific persona). hermes reads the voice-dna file to write copy that actually sounds like your brand. Ad generation and page builds all read from the psychographic map and personas that brand-context produces.