Building and Maintaining Your Brand Kit with MCP

Your Brand Kit only works if every AI tool in your stack can read and follow it. Rafaël shows you how to connect your Brand Kit through MCP (Model Context Protocol) and enrich it in minutes so your guidelines drive every piece of content your system produces.

What You'll Learn

TL;DR

  • MCP lets you manage your Brand Kit through conversation with Claude instead of copying from docs into form fields.
  • A gap analysis shows you exactly what's missing or thin in your Brand Kit before you start enriching it.
  • You can add content types, audiences, and visual guidelines in minutes by pointing Claude at existing source material.
  • Scoped writing rules let each content type and audience carry its own guidelines on top of your global rules.
  • A complete Brand Kit powers on-brand output across Workflows, Grids, Power Agents, and any MCP-connected tool.

What MCP changes about Brand Kit maintenance

MCP (Model Context Protocol) lets you manage your Brand Kit through conversation. You iterate with Claude, which reads and writes to your Brand Kit directly, so you're never copying from docs into form fields.

In this lesson, Rafaël walks through adding a content type, an audience, writing rules, and visual guidelines, then ties everything together by generating a branded AEO (Answer Engine Optimization) dashboard.

Setting up the MCP connection

If you're on Claude Web or Desktop, connecting to AirOps is a one-click setup through the connector directory. Once connected, you can see all the available tools.

Rafaël recommends downloading the Brand Kit Editor skill to your Claude workspace. It's optional, but the skill guides Claude through reviewing, filling, and updating your Brand Kit step by step, with a dedicated review UI before any changes go live.

Running a Brand Kit gap analysis

Start by asking Claude to review your Brand Kit and tell you what's incomplete. Claude loads the full Brand Kit and returns a structured audit: what's well-defined, what's missing, and what could be improved.

In Rafaël's example, the audit flagged thin content samples, a blog post content type with no template, duplicate global writing rules, and audiences and regions missing descriptions and scoped writing rules.

This gap analysis gives you a clear starting point for where to focus your effort next.

Adding a content type with a live sample

You can create a new content type by pointing Claude at a live URL as a reference. Rafaël names a new content type "how-to guide" and provides a sample article URL.

Claude reads the article and proposes a plan: create the content type, use the sample as a reference, and define a template outline for producing future how-to guides.

A built-in review UI in Claude lets you accept all changes, reject all changes, or reject specific fields before publishing.

Each content type can carry its own scoped writing rules that sit on top of your global rules. A blog post might allow casual phrasing, while a legal page would not.

Building out an audience definition

Audiences shape how the same product sounds depending on who you're talking to. In a previous lesson, you may have created a placeholder. Now you can flesh it out using context that already exists: a persona doc, a PDF, a Notion page, or a Gong summary.

Rafaël adds a new audience called "agencies" using information pulled from his website. Claude reads the source material and writes a structured audience definition covering what they care about and how to speak to them.

Once the audience exists, you can add writing rules scoped directly to it, so your content adapts automatically when that audience is selected.

Adding visual guidelines

Visual guidelines give AI your color palette, typography, logo variants, and visual rules. Without them, AI-generated assets look generic. With them, outputs look like they came from your brand team.

  • Claude can pull an entire visual identity from a public style guide within minutes: color palettes, typography rules, and logo specifications.
  • If your style guide is private, you can drop PDFs or documents directly into Claude for the same result.

Putting it all together with an AEO dashboard

At the end of the lesson, Rafaël generates an AEO (Answer Engine Optimization) performance snapshot showing mention rate against competitors. The dashboard pulls analytics data and styles itself automatically using the Brand Kit's color palette and typography.

This is a practical demonstration of what a complete Brand Kit enables: structured brand context flowing into every downstream tool without manual formatting or design work.

Why this matters going forward

A Brand Kit is the foundation that Playbooks, Grids, Power Agents, and MCP-connected tools reference automatically. Updates propagate everywhere without re-briefing individual tools.

The next lesson in the course covers connecting your site map, Google Drive, and Gong transcripts to build a Knowledge Base, which is the other half of the context that powers your content system.

Key takeaways

  1. Your Brand Kit has gaps you can't seeClaude loads your full Brand Kit and returns a structured audit. In Rafaël's case, it flagged duplicate global writing rules, a content type with no template, and audiences missing scoped rules. You get a prioritized fix list before you write a single word.
  2. Publish without leaving the chatAfter reviewing proposed changes in Claude's built-in UI, you can accept or reject individual fields and publish the updated Brand Kit directly through MCP, so you don't need to switch back to the AirOps app to finalize.
  3. Private docs work as fast as public onesYou don't need a public style guide to populate visual guidelines. Drop a PDF, a Notion page, or a Gong transcript directly into Claude and it extracts the same structured data. Custom brand fonts are the only piece that still requires a manual upload.
  4. The Brand Kit Editor skill adds a safety netIt's an optional download that walks Claude through reviewing, filling, and updating your Brand Kit step by step. The dedicated review UI means nothing changes in your Brand Kit until you've approved it, which reduces the risk of unintended updates when iterating quickly.
  5. One AEO dashboard proved the whole systemRafaël generated a competitive AEO performance snapshot styled entirely from his Brand Kit's visual guidelines. Claude pulled analytics data and applied brand colors and typography automatically, turning raw data into a polished, shareable report without manual design.

FAQs

MCP (Model Context Protocol) is an open standard that lets AI models read from and write to external systems directly. When connected to a Brand Kit, MCP allows tools like Claude to load your full brand identity, propose updates, and publish changes without switching between apps. Instead of copying guidelines from documents into form fields, you iterate through conversation, and the AI enters the data for you.

Resources

  1. Brand Kits documentationDoc
  2. Visual Guidelines documentationDoc
  3. AirOps Brand KitArticle