Keeper Security is making its secrets management platform more accessible to artificial intelligence (AI) agents by adding support for the Model Context Protocol (MCP).

Developed by Antropic, MCP has emerged as a de facto standard for enabling AI applications and agents to access data that Keeper Security is embracing via a Model Context Protocol (MCP) AI Agent Integration for Keeper Secrets Manager.

Jeremy London, director of engineering for AI and threat analytics for Keeper Security, said MCP support will make it possible to via a natural language interface query the Keeper Security platform to determine which end users and applications, including AI agents, are accessing which secrets. Keeper Secrets Manager users can now, for example, allow AI assistants to perform tasks such as generating secure passwords, retrieving and updating secrets, managing file attachments and running system health checks.

The integration will also reduce the overall amount of custom coding that would be required to programmatically launch those queries, he added. End users of the Keeper Security platform will also no longer have to copy and paste information, which increases efficiency and reduces memory-based risks.

In general, the goal is to make a platform for managing secrets, such as passwords, a lot more accessible to both cybersecurity and IT operations teams, noted London.

Over time, the number of AI agents accessing the Keeper Security platform for managing secrets is about to exponentially increase, noted London. The integration with MCP provides a mechanism to enable AI agents to only be able to access specific designated folders to enable cybersecurity teams to enforce least access privileges, he added.

There is little doubt at this point that AI agents, in addition to being pervasively employed across highly distributed IT environments, are also going to become critical tools for cybersecurity teams. In fact, the ongoing critical shortage of skills that has plagued cybersecurity teams for decades now might soon be greatly mitigated by AI agents that are assigned various tasks to complete. It’s not likely AI agents will eliminate the need for cybersecurity professionals any time soon, but the overall amount of tedium that they regularly encounter will be steadily reduced.

It may be a while yet before MCP servers are ubiquitously incorporated into security operations, but as AI agents are deployed, there may even come a day when the proverbial cybersecurity playing field becomes much more level than it is today. Of course, cybercriminals will also have access to AI agents that they will use to create more cyberattacks than ever, but it’s already apparent that AI will provide defenders with the capabilities required to both identify and mitigate threats in near real time. In effect, cybersecurity teams are now caught up in an AI cybersecurity arms race.

In the meantime, however, the overall size of the attack surface in the age of AI is also going to expand, so cybersecurity teams would be well advised to develop a strategy now to minimize the number of breaches involving AI agents and other related technologies that at this point are all but inevitable.