THE BRAND BRAIN IN OTHER AI TOOLS

Carry the Brand Brain into the AI tools you already use.

Levrit provides a read-only MCP endpoint on every paid plan. Any AI tool that supports MCP can read the Brand Brain, one campaign, evidence from reports, and the latest report in the client workspace you choose. Nothing is written, generated, or sent from outside Levrit.

FOUR TOOLS, ALL OF THEM READ

What can be read from outside.

Four tools: get_brand_brain reads the active workspace's Brand Brain (voice, audience, offers, banned phrases); get_campaign reads one campaign by its id; search_evidence searches the traceable figures in saved reports; draft_report returns the latest report as context. None of them generates or changes anything.

ONE TOKEN PER WORKSPACE

Tokens are made on the Brand page, per client.

On the Brand page, a workspace owner creates a token with a label, copies it once, and uses it as a bearer against Levrit's MCP endpoint. Each token is bound to one workspace, so an outside tool sees only the client whose token you handed it. A token can be revoked at any time from the same page.

WHAT IT CANNOT DO

Read, yes. Write, no.

The endpoint does not generate content, store output, publish, or send. Everything that leaves Levrit still goes through the stages you approve inside the product. An outside tool gets context, not control.

QUESTIONS

Can an outside AI tool create content through Levrit's MCP?
No. All four tools only read: the Brand Brain, one campaign, evidence from reports, and the latest report. Content is still made inside Levrit, through the stages you approve.
Which plans have MCP access?
Every paid plan. A workspace owner creates the token on the Brand page, copies it once, and can revoke it at any time from the same page.
If I hold five clients, what does an outside tool see?
Only the client whose token you handed it. Each token is bound to one workspace, so an outside tool never sees another client.
Create a token on the Brand page