Celigo Launches AI Agent Access Controls, Sets November Registry

Celigo made its Model Context Protocol (MCP) Management platform generally available Oct. 20, giving businesses a way to limit which tools AI assistants and agents can use and trace their actions to verified users.

The platform’s gateway applies access rules to individual tools on MCP servers built by Celigo and other vendors. It also records which tools ran, giving administrators a record of activity tied to the user behind each request.

That tool-level control lets a company permit a specific action without opening every tool available on the same server. Companies can also create tools for particular tasks using Celigo’s more than 1,000 app connections, including connections to systems that lack their own MCP servers.

MCP is an open standard that connects AI applications to external data and tools. Anthropic introduced it in November 2024. Harvey’s MCP policy engine for legal workflows is another approach to controlling access to those tools.

“Enterprises need to know who is behind every AI request, what it is authorized to do and what actions it has taken,” Celigo Chief Product Officer Matt Graney said.

Celigo plans to make its MCP Registry available in November. The registry is designed to show which servers an organization uses, which AI assistants and agents connect to them, and who has access.