Specialized agents and tools are distributed across teams, functions, and ecosystems causing agent sprawl. Without a scalable way to find and track these agents, you risk creating a black box of ungoverned agents, trapped in silos, leading to one and done solutions.
Agent Scannerspart of MuleSoft Agent Fabric, automatically find and catalog AI agents from providers like Salesforce Agentforce, Amazon Bedrock, Google Vertex AI, and Microsoft Copilot Studio into MuleSoft Agent Registry, eliminating manual data entry.
Why should you have all of your agents in one registry?
When you have all your agents in a single registry, discovering your agents for reuse is just the beginning. Centralization transforms a fragmented AI landscape into a collaborative ecosystem. It provides the complete visibility needed to track agent performance in real-time and the granular control required to govern security and access from a single, unified plane.
Unified governance and security
When agents are scattered across different platforms, it is nearly impossible to enforce a single security standard. With the agents cataloged in Agent Registry, you can now apply a single set of enterprise policies (like PII data detecting or access control) through Anypoint Flex Gateway.
Enable Agent-to-Agent (A2A) collaboration
The most powerful AI workflows involve agents talking to other agents to solve complex business tasks. For example, a customer support agent might need to collaborate with a Shipping Agent to track a package.
Furthermore, you can use MuleSoft Agent Broker to intelligently route requests to the appropriate agent cataloged in the registry. Instead of one to one connections, you are now able to create a network of agents to effortlessly manage and direct diverse AI agents for complex, secure, and high-quality process execution.
Agile tech stack
Modern AI moves fast. You might switch platforms to leverage a model better suited for a specific language, reduce latency, or take advantage of more competitive pricing. The key is to not inhibit your developers but still having a standardization that provides visibility and governance regardless of where the agent lives.
Real-time observability
A centralized registry coupled with monitoring and visualization turns agent sprawl into a visual map using MuleSoft Agent Visualizer. With the agents traceable through a single control plane using MuleSoft Agent Fabric, you are able to track which agents are used most, which ones are reused, and where the potential bottlenecks are.
How to use Agent Scanners
Now that we’ve discussed why you would want to use Agent Scanners to create a single repository of agents, let’s dive into how you can set one up to scan for new and updated agents in Amazon Bedrock.
Prerequisites
You will need the following to set up an Agent Scanner:
- Exchange Administrators permission for your user
- Verify that you are in the business group where you want to add the scanner
- Amazon Web Services Access Key ID and Secret with following permissions
- bedrock:ListAgents
- bedrock:GetAgent
- bedrock:ListAgentAliases
- bedrock:GetAgentAlias
- bedrock:ListAgentVersions
- bedrock:GetAgentVersion
- Agents prepared and deployed with an alias in Amazon Bedrock
- As an example for this guide, we have five agents deployed in Amazon Bedrock
Create a new scanner in Anypoint Platform
Follow the steps below to set up a new scanner that can discover and catalog agents in Amazon Bedrock.
- Log in Anypoint Platform
- Go to Exchange and click on Agent Scanners tab
- Enter the scanner name. For example: Amazon Bedrock Scanner.
- Enter in a description (Optional)
- Configure Scanner Run Schedule depending on how often you want the scanner to run
- Decide if you want Syncs to be auto-resolved or be asked
- Under Agent Providerchoose Amazon Bedrock
- Input in Access key information and AWS Region that your agents are deployed in
- Verify authentication by clicking on Test Connection
- You can configure email notifications to get the scan summaries in Advanced Settings (Optional)
- Click Add Scanner
Let the scanner do the boring work for you
Now that you have a scanner created, click on Run now to import all the agents into Amazon Bedrock.
Once the scan is done, you can get a summary of how many agents were added as well as the scan history.
In Exchange, you can now see the five agents that were imported.
When you click on one of the agents, you can take a look at the metadata that was imported such as the A2A agent card.
Automated discovery and centralized management with Agent Scanners
Cataloging your AI agents shouldn’t be a manual, time-consuming task. By using Agent Scanners, you automate the discovery process across different cloud providers, ensuring your Agent Registry is always accurate without the risk of human error.
Centralizing your agents is the first step toward effective management. Once they are in a single repository, you can consistently apply security policies, monitor usage, and enable different agents to work together. Instead of managing a collection of disconnected agents, you gain a clear, organized view of your AI resources, allowing your team to focus on building rather than tracking down data.
To learn more about Agent Scanners, watch our automated discovery demo. And don’t forget to check out our interactive demo, Start your Agentic Transformation with MuleSoft, to see how Agent Scanners fits with your agentic transformation needs!
