# brand.md: Brand Guidelines Agents Can Actually Read

> Canonical: https://auxfirst.com/news/brand-md-context-file.html
> Markdown representation of this article, for agents and LLM crawlers.

Your brand book is a PDF and the systems writing your copy can't parse it. The brand.md pattern, the 8 sections that belong in it, and a working template.

By Emil Krzemiński, founder of auxfirst. Published 12 August 2026.

## The problem with a 90-page brand PDF

A brand team spends four months and a meaningful budget producing the book. It is genuinely good. It ships in March.

By September the social agency is generating captions with an AI tool, and someone has pasted pages 12 to 14 into its settings. The ecommerce team's product-description generator has a different excerpt, chosen by a developer who needed something in a hurry. Customer service has a support assistant whose tone was written from memory. The paid team's creative testing tool has nothing at all.

Four brands in production, all drifting away from the book and from each other, and no one positioned to notice. This is not a governance failure. It is a format failure: the brand system was never expressed in a form a machine could consume, so every machine got a hand-cut fragment of it.

## What a brand.md is

A single Markdown file, ideally under 300 lines, holding the parts of your brand system that a machine needs in order to produce or check on-brand output.

The constraint is the point. This is **not the brand book in Markdown**. It is the operating subset: the rules that have to be applied every time something is written, stated in a form that can be applied without interpretation.

If your first draft is 900 lines, you have made a copy of the book. Cut it until every remaining line would change an output if it were removed. There is now evidence from coding agents that long context files raise cost without raising quality, and that anything named in one gets reached for whether or not it should be.

## Why Markdown and not PDF, Word or Figma

- **It diffs.** You can see exactly what changed and who changed it. A PDF gives you two PDFs.
- **There is no extraction step.** Parsing a PDF loses column order, table structure, anything inside a figure. What you reviewed is not what arrives.
- **It is reviewable by non-technical people.** Unlike JSON or YAML, a brand lead can read and edit it without training — which is what makes ownership realistic.
- **It is what the agent ecosystem converged on.** The tools that read context files read Markdown, without exception.

**One claim not to make.** This does not help your search or AI-answer visibility. Google's guidance states that you don't need machine-readable files or Markdown to appear in Search or its generative features, "as Google Search itself doesn't use them." The case for brand.md is about feeding agents you operate. That is narrower than the claim usually made, and it has the advantage of being true and testable.

## What goes in a brand.md

Eight sections, in roughly this order, because what comes early carries more weight.

1. **Who we are** — one paragraph, written to be quoted verbatim. Not a mission. The answer you want an assistant to give.
2. **What we sell** — plain nouns. "A yoghurt drink" beats "a wellness ritual". Vagueness compounds through a generation step in a way it doesn't through a human writer, who would ask.
3. **Voice** — three adjectives, each with a sentence of anti-pattern. "Warm, but not chummy. We don't use exclamation marks or address the reader as 'friend'." That is enforceable. "Warm" is not.
4. **Words we use / words we don't** — a literal two-column list, ten to thirty pairs. The highest-leverage section in the file, because word choice is the thing generation gets wrong first and reviewers catch last.
5. **Claims we can make** — each with its substantiation status and expiry attached.
6. **Claims we cannot make** — regulatory and legal lines as hard prohibitions. "Never state or imply X", not "be careful with X". An agent resolves "be careful" in whichever direction the rest of the prompt pushes it.
7. **Names, spellings, capitalisation** — boring, and constantly wrong without it. Stops the small errors that make output visibly machine-made.
8. **Escalate, don't answer** — pricing, legal, active incidents, competitors by name, journalists. The section that makes the file safe to deploy, and the one most often missing. An agent with no escalation list will attempt everything.

Example of section 4:

| We say | We don't say |
|---|---|
| customers | users, consumers |
| helps with | cures, treats, fixes |
| plant-based | vegan-friendly, guilt-free |

## The template

Download: https://auxfirst.com/downloads/brand.md — no email required.

```
# brand.md — [Company]
Last reviewed: [date] · Owner: [name, role]

## Who we are
[One paragraph, written to be quoted verbatim.]

## What we sell
- [Product]: [what it physically is, in plain nouns]

## Voice
- [Adjective] — but not [anti-pattern]. [One concrete rule.]

## Words we use / words we don't
| We say | We don't say |
|---|---|
| [term] | [term], [term] |

## Claims we can make
- "[claim]" — substantiated by [source], valid until [date]

## Claims we cannot make
- Never state or imply [claim]. [Reason, one clause.]

## Names and spellings
- [Product name] on first use, [short form] after.

## Escalate, don't answer
- Pricing → [team]
- Legal, regulatory, or safety → [team]
- Anything naming a competitor → [team]
```

## Where the file lives

Four options, in descending order of how well they hold up.

- **Version control, with a published URL.** History, review, and an address you can point any tool at.
- **A shared drive, single canonical file.** Workable. Weak on change history, and edit permissions are usually far broader than anyone realises.
- **Pasted into each tool's system prompt.** Fast, and guarantees drift.
- **Inside the brand book as an appendix.** Feels tidy. Nothing will ever read it.

Recommendation: version-controlled, with a stable URL, and every tool pointed at that one address rather than given a copy.

## Who owns it and how it stays current

One named owner in brand — a person, not a team. One named reviewer in legal, for the two claims sections only, so review doesn't bottleneck voice edits. A review trigger tied to campaign cycles and product launches rather than a calendar reminder.

**How you know it's working.** Keep a set of ten prompts that have historically produced off-brand output. Run them before the file and after it, and again after every meaningful edit. Two things fall out: evidence the file is doing something, which is what makes it fundable past the first enthusiasm; and early warning when a well-intentioned addition makes output worse — which happens more often than people expect, because everything in the file competes for attention with everything else in it.

Ten prompts, a spreadsheet, twenty minutes a quarter. If nobody will do that, nobody owns the file.

The failure mode to watch for is the file written once, read by six systems, and not opened by a human in five months. That is worse than not having one, because everything downstream is now confidently consistent with a document nobody has checked.

## What this is not

- **Not a prompt library.** It describes the brand; the prompt describes the task.
- **Not a replacement for the brand book.** The book keeps strategy, rationale and the visual system.
- **Not an SEO or AI-visibility asset.** It is read by the agents you operate, not by search engines.

The adjacent file — governing how work gets *done* rather than how it should *sound* — is AGENTS.md. The two are siblings, and most teams need both before either works properly.

## Read next

- [AGENTS.md for teams that don't write code](https://auxfirst.com/news/agents-md-for-non-engineering-teams.html)
- [Prompt injection in .md context files](https://auxfirst.com/news/markdown-context-file-security.html)
- [llms.txt: what 137,000 sites' server logs actually show](https://auxfirst.com/news/llms-txt-evidence.html)
- [MD files explained — the complete .md file format guide](https://auxfirst.com/news/md-file-markdown-guide.html)
- [The context file checker](https://auxfirst.com/tools/context-file-checker.html)

## Sources

- Google Search Central, "Optimizing your website for generative AI features on Google Search" — https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Gloaguen et al., "Evaluating AGENTS.md", arXiv:2602.11988, February 2026 — https://arxiv.org/abs/2602.11988

Method: the brand.md pattern is auxfirst's own; the eight-section structure comes from client work rather than from a specification, and we have said so rather than implying a standard exists.

---

Emil Krzemiński is the founder of auxfirst (https://auxfirst.com/), the agentic experience design agency.