Salesforce-owned integration platform provider MuleSoft has added a new feature called Agent Scanners to Agent Fabric — a suite of capabilities and tools that the company launched last year to rein in the growing challenge of agent sprawl across enterprises.

Agent sprawl, often a result of enterprises and their technology teams adopting multiple agentic products, can lead to the fragmentation of agents, turning their workflows redundant or siloed across teams and platforms.

This fragmentation more often than not undermines operational efficiency and complicates governance, making it significantly harder for enterprises to scale AI safely and responsibly.

MuleSoft’s Agent Scanners is designed to help enterprises discover agents across diverse environments, such as Copilot, Vertex AI, Bedrock, and Agentforce.

After detecting the new agents, the scanners automatically synchronize the agents and their metadata with Agent Registry, which in turn can be used to make the agents discoverable by other agents or developers, the company said.

Agent Registry is one of the many tools inside Agent Fabric, and it acts as the centralized directory that catalogues the capabilities, metadata, and endpoints of both in-house and third-party agents. Other tools inside Agent Fabric include Agent Broker, Agent Visualizer, and Agent Governance.

Simple yet foundational for CIOs

Agent Scanners’ automated agent discovery ability, although deceptively simple, could be more foundational for CIOs trying to adopt agents in production use cases.

“For CIOs and security leaders, the biggest issue today isn’t deploying agents, it’s understanding what’s already running. Many enterprises struggle to answer basic questions like how many agents exist, where they’re deployed, which models they use, and what data they can access. Agent Scanner directly tackles that visibility gap,” said Robert Kramer, principal analyst at Moor Insights and Strategy.

CIOs can also use the tool to cut down on fragmentation, Kramer noted, adding that teams in enterprises often build agents in isolation, duplicating effort because they don’t know what already exists elsewhere internally.

Agent Scanners’ significance, however, extends beyond its appeal to CIOs and security leaders, particularly as the pace of agent creation accelerates across large enterprises.

Stephanie Walter, practice leader for the AI stack at HyperFRAME Research, said the capability is also foundational to the original goal of Agent Fabric itself: curbing agent sprawl rather than just documenting it.

“Manual tracking is already failing. I expect so many agents to be created in the near future that a manual registry would be obsolete within weeks. Without an automated scanner, an enterprise’s Agent Registry becomes a stale spreadsheet that fails to account for the shadow AI being built in various cloud silos,” Walter added.

Further, analysts said Agent Scanners’ ability to automatically read and document agents’ metadata could give CIOs a more practical way to maintain visibility as agent deployments scale, since simply knowing that an agent exists does little to support governance, cost optimization, or reuse.

“The real value comes from understanding context — what the agent can do, which LLM it uses, and what data it’s authorized to touch,” Kramer said.

“Agent Scanner’s metadata extraction adds that depth, making it easier for security teams to assess risk, for architects to identify overlap or consolidation opportunities, and for developers to safely connect agents. That shifts agent management from static inventory tracking to something that supports real operational decisions,” Kramer added.

Seconding Kramer, Walter pointed out that the same metadata can then be standardized into Agent-to-Agent (A2A) card formats, making identifying and trusting agents easier.

Agent Scanners is currently in preview, and MuleSoft expects the capability to be made generally available towards the end of this month.