Markolé MCP: Connecting AI Agents Directly to Your Brand

Markolé now speaks Model Context Protocol (MCP) — the open standard that lets AI agents like Claude Desktop, Claude Code, and other MCP-aware clients connect directly to external systems through a typed, structured interface.

In practice, this means your brand is no longer just a PDF an AI has to interpret. It becomes a live, structured source your agent can read from, and — on the Pro tier — write back to, under the same rules Markolé itself enforces.

1. What MCP Unlocks

With MCP, an external AI agent can:

Read your full brand DNA — narratives, traits, visual identity, presentations, and the complete brand book — with proper semantic labeling.
Stay perfectly on-brand in any tool you use, because it works from the same structured data Markolé uses internally.
On the Pro tier, write back to your brand: update narratives, regenerate logos, produce brand imagery, modify presentations, or even create entirely new brands autonomously.

MCP is a paid feature:

Guided users get read-only access — perfect for keeping any external agent on-brand without changing anything inside Markolé.
Pro users get the full write surface — designed specifically for agencies, brand consultants, and brand teams who want to use Markolé as their structured brand repository and benefit from Markolé’s logo, imagery, and presentation tools on top of it.

2. Enabling MCP and Creating a Token

Open Settings → MCP Access.
Toggle Enable MCP access on. This is required before any token can be used.
Click Create new token. Give it a descriptive name (for example: Claude Desktop – personal laptop).
Copy the token immediately — it starts with mrkl_pat_... and is only shown once. Store it like a password.
To revoke a token, return to Settings → MCP Access and click Revoke on any active token. Revocation is immediate.

If you ever lose a token, simply revoke it and create a new one.

3. Connecting Claude Desktop or Claude Code

In Settings → MCP Access, you’ll find a copy-ready Claude Desktop config snippet. Expand the snippet, click Copy, and paste it into your Claude Desktop configuration file. Replace the token placeholder with the token you created above and restart Claude Desktop.

The same token works for any MCP-aware client: Claude Code, custom agents built on the Anthropic SDK, or any third-party tool that supports the MCP protocol.

Once connected, your agent can immediately list your brands and begin reading brand content.

4. What Agents Can Read

Every MCP-connected agent can:

List your brands and teams you have access to.
Read brand summaries, narratives, traits, and visual identity.
Pull the full brand book or final report as JSON-LD.
Read the visual identity as JSON-LD.

This is the structured, semantically labeled version of your brand — not a flattened export. Your agent always sees the latest state.

5. What Pro Agents Can Write

On the Pro tier, agents can perform real, lock-aware writes:

Update brand narratives — vision, mission, values, archetype, or any of the 17 narratives.
Update visual identity — color palettes, typography, dos & don’ts.
Upload or generate logos — drop in an existing logo from a URL, or have Markolé generate logomark, logotype, and dark variants with stored creative briefs for future regenerations.
Generate brand imagery — produce primary 3:2 brand images using full brand context (existing logos, imagery, and product references all feed in).
Update brand presentations — replace a presentation entirely; slide background images are regenerated automatically when the brief changes, and old assets are deprecated cleanly.
Update the final brand and visual reports — write back synthesized brand content.
Create new brands — Pro agents can autonomously create new Markolé brands in any team where you’re an admin or editor, defaulting to your personal team.

6. Section Locks and Safe Writes

Every write respects the section locks you set inside Markolé.

If a section is locked, an agent that tries to modify it receives a structured error listing the exact locked sections plus clear unlock guidance.
This prevents silent overwrites — your agent always knows what it can and cannot change.
For more on managing locks, see Locking and Unlocking Brand Sections.

Updates from MCP also broadcast in real time: connected teammates see the changes immediately with the usual lock indicators and refresh states.

7. Why This Matters for Agencies and Consultants

If you manage brand identity for multiple clients, MCP changes how you work.

Point your agent at Markolé as the single source of truth for each client.
The agent reads structured brand content, writes back updates, and generates production-ready logos and imagery — all without copy-pasting briefs or juggling PDFs.
Each client sees every change live in the same workspace, with full audit and lock control on their side.

8. Tips and Limits

MCP must be enabled for tokens to work — even valid tokens are rejected when the toggle is off.
Free users do not have MCP access; upgrading to Guided or Pro unlocks it.
Treat tokens like passwords. Revoke unused tokens regularly.
Realtime sync is on by default — any teammate connected to Markolé will see MCP-driven updates appear live.

For a complete reference of the MCP tools, resources, and the open JSON-LD brand schema, see Exporting Your Brand as JSON-LD.