翻訳待ち：Manage agents, tools and skills at scale with AWS Agent Registry

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

Most organizations scaling their use of agents and tools hit the same challenges. Teams build in isolation, with no shared record of what exists, who owns it, or whether it’s been reviewed. The problem has moved from building agents and tools to discovering and governing them. AWS Agent Registry is purpose-built to solve this. Now generally available, it gives teams a single, searchable, governed catalog for agents, tools, skills and custom resources in their environment. In this post, we cover what Registry is, walk through its core publishing, curation, and discovery workflows, explore enterprise considerations, and look ahead at what’s next. Why do enterprises need a registry As organizations scale their agentic AI systems, three key challenges emerge. No authoritative inventory When every team maintains their own collection of agents and tools in isolation, there’s no single record of what exists, who owns it, or whether it’s still actively maintained. The result is duplicative effort, version drift, and an ever-growing sprawl of untracked capabilities scattered across the organization. No cross-team discovery Even when great tools, agents, and skills exist, they go unused because developers on other teams can’t find them. Without a searchable catalog, the default is to rebuild what already exists. That’s wasted engineering effort, redundant infrastructure to maintain, and ongoing operational cost that scales with every team that builds in isolation. No governance or audit trail Without a central registry, there’s no way to track who has access to which agents, tools, and skills, whether they’ve passed security review, or how to trace failure back to specific version and owner. At enterprise scale, each registered agent, tool, or skill needs an owner, a clear lineage, and an audit trail. AWS Agent Registry addresses all three: it provides a central, searchable catalog where teams register agents, tools and skills, discover what already exists across the organization, and apply governance through access control, lifecycle tracking, and approval workflows. What is AWS Agent Registry As an agentic system grows from a handful of tools to hundreds, finding what already exists and trusting what you find becomes the bottleneck. AWS Agent Registry alleviates that bottleneck. It gives teams a single, governed catalog to register, discover, and manage AI agents, tools, and capabilities with built-in semantic search and access control. Internally, Registry operates across two complementary planes: The Governance Plane. The Governance Plane is the comprehensive agents, tools, and skills that are registered. It’s designed to be the authoritative store for resources within its defined scope, regardless of their lifecycle state. This is where admins configure the rules that shape how resources are managed, such as: Compliance and security signals – metadata that tracks whether a resource has passed security review, meets regulatory requirements, or carries known risks. Discovery policies – entitlement-based search rules that control which consumers can see which resources based on their role or team. Custom metadata schemas – organization-specific fields (for example, cost center, data classification, SLA tier) that standardize how teams describe their resources. Over time, the Governance Plane will surface richer governance signals, giving admins a single view into the compliance and security posture of their entire agentic landscape. The Discovery Plane. The Discovery Plane is what consumers interact with day to day. It presents a curated, high-performance view of only the resources that have passed the organization’s approval bar. Key characteristics: Curated, not comprehensive – only records approved by an admin or curator appear here. Draft, rejected, or shadow resources are not visible to consumers. Built for scale – supports high-throughput queries so agents and developers can search programmatically without hitting rate limits. Semantic and lexical search – consumers find resources by intent (“find me a tool for ticket routing”) or by exact name, across the full approved catalog. Trust signals, not governance details – surfaces summarized compliance and security indicators framed to make decisions of using it rather than raw governance data. Together, the two planes separate concerns cleanly: admins get comprehensive visibility and policy control through the Governance Plane, while consumers get a fast, governed search experience through the Discovery Plane that only shows resources ready for use. Some of the features described earlier are forward looking and are detailed in a later section. What can be cataloged in the Registry Registry supports four record types: MCP – Model Context Protocol server, its tools, resources, and prompts. Agent – Agent2Agent (A2A) agent card defining agents and their skills. Skill – agent skill definitions in markdown files and associated code/packages. Custom – Custom descriptor which must be valid JSON. The following diagram gives an overview of Registry, its capabilities, and access surfaces. Figure 1: Overview of AWS Agent Registry and its access surfaces What customers are saying Customers and Partners across industries, segments, and geographies are already seeing the value of the AWS Agent Registry for solving the discoverability, governance, and operational challenges that emerge as organizations scale to managing hundreds or thousands of AI agents. Companies like Sony are using it to reuse agent patterns across business units. Mitsubishi Electric are looking to give developers a single place to discover and trust what they build against. For companies like Southwest having a centralized way to discover and govern agents helps teams move faster with confidence, reducing redundancy, preventing sprawl, and scaling innovation in a way that’s sustainable for the enterprise. “At Southwest Airlines, we’re building agentic AI tools and autonomous agents to streamline operations across our 70,000+ employees and enhance the experience for the millions of Customers we serve every day. We went from dozens of agents and tools scattered across multiple technology teams with no shared record of what existed to a single, governed catalog that the entire organization trusts. AWS Agent Registry gives our platform team complete visibility into what’s deployed, who owns it, and whether it’s been reviewed. Our developers now find approved capabilities through semantic search in seconds instead of rebuilding what another team already built. Registry cut duplicative development effort significantly and became the backbone of how we govern agentic AI.” — Lauren Woods, CIO/EVP, Southwest Airlines And for customers like PepsiCo, having a system of record for discovery and usage becomes critical. “Agent Registry solves that. It gives teams a centralized way to discover, govern, and reuse agents, tools, and integrations across the enterprise. Customers like PepsiCo are already thinking about this at scale.” — Athina Kanioura, Chief Strategy & Transformation Officer, PepsiCo At Syngenta, the Registry became the foundation of their AI agent governance strategy, providing a centralized catalog that eliminates redundant rebuilding across teams. “At Syngenta, we built our AI agent governance on AWS Agent Registry. It gives our teams a single, trusted catalog of AI agents, tools, and skills. Teams publish once, then discover and reuse what already works – instead of rebuilding agents, connectors, and business procedures from scratch. We register, review, and approve every capability before sharing it across the organization, retaining clear ownership, versioning, and control over security and access.” — Sandeep Rayasa, Enterprise Architect, Data and AI “As a leading software and AI solutions to telecom industry, Amdocs is an early adopter of AWS Agent Registry. By integrating it into our aOS Cognitive Core platform, we gain a unified view of agent assets across diverse environments while streamlining governance, compliance, and lifecycle management. The Registry’s framework-agnostic design aligns with our open platform strategy and enables us to deliver a trusted control plane for managing large-scale agent ecosystems, accelerating AI adoption across the telecom industry.” — Ron Dublero, Cognitive Core, Chief Software Architect, Amdocs AWS Partner perspectives As the number of agents grow across multiple cloud environments, managing their lifecycle, preventing duplication, and governance becomes operationally critical. With teams building agents across AWS, the Registry is the centralized governance layer that standardizes how agent records are shared, discovered, and governed. “We see great value with AWS Agent Registry at our enterprise clients. It manages agent skills consistently across systems, while allowing for customized integration with their existing technology landscape. Beyond accelerating adoption, the centralized registry helps mitigate a common, emerging operational risk at scale: agent sprawl.” — Dr. Binqi Zhang, Managing Director, PwC Australia “As agents proliferate across an organization, it gets harder for humans, agents, and tools to know which agent to use when. AWS Agent Registry removes that undifferentiated heavy lifting, giving the right context to the right agent at the right time, and provides a scalable, secure way to grow as more agents get deployed. At Caylent, we worked with customers through the beta and deployed it in our own environment to cut the engineering friction of delivering growing multi-agent architectures.” — Randall Hunt, CTO, Caylent “AWS Agent Registry gives our clients the missing piece of enterprise agent governance: a single source of truth for discovering, authenticating, and trusting agents across the enterprise. This is a key capability to enable multi-agent architectures at scale, interconnecting line of business units.” — Pinaki Karfa, AI architect, Slalom Partners and ISVs are extending the AWS Agent Registry by building integrations that bring MCP tools and security directly into agentic workflows. From grounding agents in trusted data to continuously assessing security posture, the registry becomes the single source for not only where agents are discoverable, but also tools and skills that are secure for use at scale. Informatica is building the integration for Informatica hosted MCP (Model Context Protocol) servers to be listed, discovered, and securely integrated with agents through the AWS Agent Registry. The MCPs extend Informatica’s data management capabilities (including metadata exploration, data quality, and master data management) directly into agentic AI workflows. This provides the necessary data foundation for AI agents to access and act upon trusted, governed data supporting accuracy and compliance across the enterprise. “Informatica from Salesforce is proud to be a launch partner for the AWS Agent Registry. With Informatica MCP servers now discoverable in the Registry, enterprises can easily activate trusted, governed data and extend the full power of IDMC into agentic workflows on AWS. This means AI agents that are not just intelligent but also grounded in high-quality data, driving accuracy, trust, and real business impact.” — Gopinath Sankaran, VP, Strategic Cloud Alliance, Informatica from Salesforce Check Point is integrating Agent Registry to give customers continuous discovery and security posture assessment of their Amazon Bedrock AgentCore workloads. “Check Point views Registry as a foundational component of AI runtime security, it gives us the intended security posture of every deployed agent. By combining Registry metadata with runtime telemetry, we can correlate what an agent was designed to do with what it’s actually doing during execution. This lets us make security decisions based not j [truncated for AI cost control]