Skip to content
Monopea

Monopea Catalog

Every tool you connect — gated by default.

Connect any MCP-compatible server, or build a tool library from API documentation. New tools start needing approval, and credentials are referenced as {{secret:NAME}} — never shown to the model.

Start freeSee pricing
crmMCP serverbillingMCP serverticketsMCP serveryour-apiMCP servercrm__update_dealgatedbilling__refundgatedtickets__closegated

MCP

Real file format, in and out

—

A surface, not a tool family

1

Approval gate for every outward action

CH · EU

Stored in Switzerland. Processed in the EU.

The Catalog, showing a grid of integrations in which only a handful read Ready and the rest await a key or a connection — read-only calls run automatically, anything outward stays review-gated.
Catalog · what the agent can already reach, and what is still waiting on a key

What you ask → what actually happens

Catalog, in the words people type.

Mechanism, not adjectives: each line is what the agent does when you say it.

“Connect our CRM.”

Adds its MCP server; each new tool arrives needing approval until you decide otherwise.

“Here are the API docs — make me tools.”

Turns the documentation into a tool library of your own, publishable to the workspace.

“Use my API key for it.”

The key goes into the write-only vault; the agent only ever sees {{secret:NAME}}.

Capabilities

What Catalog does.

01

Any MCP server

If it speaks MCP, it plugs in.

02

Fail closed

Unknown tools and new tools go to review, not straight to dispatch.

03

Build your own

API documentation in, a tool library out, served over MCP on revocable tokens.

04

Write-only vault

Secrets are PUT once and never read back — not by you, not by the model.

Tools the agent calls

Real functions, not a chat reply.

Catalog tools are whatever you connect — named <service>__<tool> and gated like every built-in.

Catalog tools are whatever you connect — named <service>__<tool> and gated like every built-in.

Governed

It works on its own. It asks at the door.

Inside the workspace the agent just works. Anything above the one risk threshold you set — and anything that leaves — becomes a proposal with its exact arguments, and waits for you.

How the approval gate works
Proposal · send renewal quotedrafted in the workspacecredentials by {{secret}} onlyabove your threshold → askwaiting…Approveactivity logdispatched

Works with

The rest of the suite, one agent.

Agent flows

A canvas for per-project agent teams: who runs, who supervises, and what each one may read. Every sub-agent inherits the same gate, and a stop cascades to everything it delegated.

Explore Agent flows

Library

Files, long-term memory and a knowledge graph the agent builds as it reads — consulted on every turn, with the evidence behind an answer graded so you can see how well-supported it is.

Explore Library

Teams

Members, roles and per-project scope, and agent personas with their own brief — the agent can run the workspace admin it is authorised for, and changes to who sees what are proposals.

Explore Teams

FAQ

Catalog, answered.

Can I connect my own MCP server?
Yes — any MCP-compatible server. Its tools start gated.
Does the model see my API keys?
No. Keys live in a write-only vault and are substituted at dispatch; the model only sees {{secret:NAME}} references.
Does the agent need my approval for everything?
No — that would make it a form, not an agent. Work inside the workspace runs on its own; anything above the one risk threshold you set, and anything that leaves the workspace, becomes a proposal with its exact arguments and waits for you.

Hand Catalog its first job.

Open a workspace, give the agent a task, and approve what it proposes. Stored in Switzerland, processed in the EU.

Start freeSee every product