Skip to main content

Govern every tool your agents can call

MCP (Model Context Protocol) servers are the connectors agents use to act on real systems. The MCP Gateway keeps them in one curated registry and decides who can call each one, how often and at what cost.

Who calls it

  • Agents
  • Users
  • Teams

MCP Gateway

Every request, in this order:

  1. 01Check accessIs this caller allowed this action?
  2. 02Filter toolsAgents only see the tools they may use.
  3. 03Run the callWithin the connector's budget and rate.
  4. 04Record itAn audit log of every call.

MCP servers

  • Salesforce
  • SAP
  • Visor

Your own servers

  • Any MCP server, by URL

Every tool call checked first

Before a call reaches an MCP server, the gateway checks the caller's permissions, the connector's budget and its rate. Calls outside the rules never leave the gateway.

  • Permissions per user, team and agent
  • Budgets for paid connectors
  • Circuit breakers on anomalous use
MCP Gateway · tool callsLive
  1. support-agentzendesk.list_ticketsAllowed
  2. ops-botpartner-api.create_orderAllowed · 62% of budget
  3. support-agentsalesforce.export_contactsDenied · not granted
  4. research-agentweb-search.queryBreaker open · 40 calls/min
Checked before forwarding: access, budget, rate1 forwarded · 0 stopped

From one server to every team, in four steps.

1

Register

Put each MCP server in the registry once.

2

Curate

Group the tools each team needs into one set.

3

Scope

Decide who can call what, down to single actions.

4

Watch

See every call, what it returned and what it cost.

Connectors for your stack, and your region

Three groups of connectors: our own products, the enterprise systems your processes run on, and any partner service or system you work with, so workflows can be automated end to end.

First-party

Our own products, ready to connect.

  • Visor
  • BuildPad
  • Workspace
  • Feedback
  • Chocolate Factory

Enterprise systems

The systems of record your processes run on.

  • CRM
  • SAP
  • Salesforce

Local partners

Any partner service or system you work with, so workflows run end to end. Integrations you won't find elsewhere.

  • Any partner system

A curated registry, by category

Every approved MCP server in one list, organized by category, with its health, the actions it exposes and the agents that use it.

  • Organized by category
  • Health and action count per server
  • Which agents use each one

Role-based access to every server

Define which users, teams and agents can call which MCP servers, down to single actions, such as allow read and deny write. Access is set per project and can change at any time.

  • Users, teams and agents
  • Down to single actions
  • Set per project

Every tool call filtered, logged and metered.

Tool filtering

Each agent sees only the tools it is allowed to use, so it can't call what it can't see.

Tool groups

Bundle chosen tools from several servers into one set, and give that set to a team or an agent.

Audit log

Every tool call recorded: which agent, which server and action, what came back and how long it took.

Cost per tool call

See what each tool call costs next to the model spend behind it, by tool, server, team or agent.

Budgets and rate limits for paid connectors

Give each paid connector a spend budget and a call rate per team or agent, so one workflow can't run up a vendor bill.

  • Spend budget per connector
  • Call rate per team or agent

Circuit breakers

When usage looks anomalous or reaches for sensitive data, the breaker trips and cuts the connector off before the damage spreads.

  • Anomalous call volume
  • Sensitive data access

Add a server from a template or a URL

Start from a template or register any MCP server by its URL. The gateway checks the connection and loads its actions.

  • API key, OAuth or custom headers
  • Or each user's own sign-in
  • Read-only option