MCP App Store
by KeenEthics
Brandrails icon

Brandrails MCP Server for ChatGPT

By Brandrails

Overview

Brandrails keeps each brand's approved colours, type, logo, products and voice. Ask for a landing page, product page, email or social graphic and it is composed on that brand from approved facts only, then published to a view link and an edit link. Save standing rules for a brand and they apply to every design after.

How to connect

  1. ChatGPT: Open the Brandrails listing in the ChatGPT app directory and choose Connect. ChatGPT listing Brandrails exposes 11 tools here.

Tools

compile_brand_dna

ChatGPT
Call this only after prepare_brand_design returns needs_compilation. Compile its complete approved compiler brief into one committed Design DNA, submit it with the same immutable projectId, then call prepare_brand_design again before designing.

get_brand_asset

ChatGPT
Call this BEFORE any image generation runs, including the assistant's own built-in image generator, whenever the picture must show an approved packshot, product photo, or logo, and whenever you need to look at the artwork itself. Image generators cannot open links: an asset URL or brandrails.io link passed to one is ignored and the result is an unrelated picture. Pass the projectId from list_projects and the assets/<semantic-filename> path from prepare_brand_design; the picture itself comes back in this conversation as PNG or JPEG pixels. Attach each returned picture to the image generation request as a reference image, one per product, so the result shows the real artwork. Never describe an approved asset and ask a generator to recreate it. If no image generator is available in this conversation, say so and tell the user to switch to a mode that has image generation; never show an asset link or other pictures in place of the requested image. A vector logo comes back as its own markup to place verbatim, and a font or any other file that is not a picture comes back as a fetch link, both with the reason; so does an asset too large to carry.

get_design

ChatGPT
Call this before revising or extending a design that already exists, with the projectId from list_projects. Pass the design's id when you know it; omit it for that project's current design. It returns the compact editable source, the version, and how much customer feedback is still open, so a revision edits the real design instead of replacing it with a fresh one. Preserve its token fingerprint and its assets/<semantic-filename> references, then publish with publish_design and the same id.

get_reference_doc

ChatGPT
Call this with the projectId from list_projects when a deliverable touches a topic the brand has its own reference doc for, before building. Omit the topic to list every available topic. The brand's own compiled docs are brand authority and win a name collision; the platform's notes are subordinate craft reference.

get_reference_render

ChatGPT
Call this with the projectId from list_projects and a designId from list_reference_renders to look at it before composing: a live-site screenshot (an id starting "site:") shows the brand as it is today; a published design shows how it has been composed before. Study its composition, photography, type scale, spacing rhythm, tone alternation and motif placement the way a designer studies a shipped page; never copy its content or duplicate it wholesale. If the render is still pending, continue from the brand guidelines and reference docs rather than waiting.

list_projects

ChatGPT
List every Brandrails project the authenticated user is currently authorized to discover. Archived projects are excluded; a project still being set up or whose billing is past due is listed, because the user can still design on it — each row carries the role and whether publishing is allowed.

list_reference_renders

ChatGPT
Call this BEFORE composing a new design, with the projectId from list_projects. It lists screenshots of the brand's live website as it looks today (kind live-site), then rendered images of its published designs, customer exemplars first. LOOK at the live site and the one or two designs closest to what you are making with get_reference_render before writing any markup. Exemplars were chosen by the customer; other designs were made by an assistant and are evidence, not a standard — learn from their judgment, improve on them, never copy their content. A design listed as pending has no render yet and will render when requested.

prepare_brand_design

ChatGPT
Call this first for a design in one project selected from list_projects. It returns the complete approved context. If state is needs_compilation, call compile_brand_dna with the same projectId, then call this tool again before designing.

publish_design

ChatGPT
Validate and publish complete HTML to an authorized Brandrails project.

save_brand_preference

ChatGPT
Call this when the customer gives standing guidance meant for every future design in one project rather than a one-off correction — "never do this again", "stop adding that", "always prefer this" (a layout habit like an unwanted sidebar, edge marker, or divider is the classic case). Send one rule per call, in plain words, with polarity avoid or prefer, and the projectId from list_projects. The rule retires the active Design DNA, so the next prepare_brand_design returns needs_compilation and the brand recompiles with it; the customer can edit or remove it in their Brandrails dashboard. When the rule is about one kind of work, start it with that kind — "For social posts: …", "For product pages: …", "For website heroes: …", "For emails: …" — so it applies only there; a rule that starts otherwise applies to everything.

screenshot_design

ChatGPT
Call this once after publish_design returns ok === true, with the same projectId and the published design's id. It returns the design as it actually renders, so you can proof your own work the way a designer proofs a print: clipped or overflowing text, collapsed sections, unreadable contrast, broken imagery. Fix any visible defect in this session and publish again with the same id before sharing the links.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/brandrails"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/brandrails)