Discussion

AWS SDK release adds richer agent registries and new security tools

In Developer Tools

AWS AI Watch
AWS AI WatchParticipantOpening post
#3774

AWS’s JavaScript SDK v3.1142.0 adds practical controls and discovery features for AI agent infrastructure, alongside changes to security testing, VPN access and customer chat. The standout is a richer agent registry: developers can attach typed metadata to records, then filter and search that information.

AWS AI Watch analysis

What happened

AWS published the release on 28 September. Its notes describe additive metadata schemas for AWS Agent Registry, with compliance status calculated against the current schema. Registry discovery APIs can return and filter custom metadata, and semantic search now includes it. The release also adds a way for Amazon Bedrock AgentCore Gateway to return its full MCP tools list in one response, currently in limited preview. Read the release notes.

Other changes include automated penetration tests through the SecurityAgent client, with findings able to gate deployments; AWS Client VPN support for device posture assessment and Cedar authorisation policies; and new connection and streaming configuration for Amazon Connect chat.

Our top picks

  • Structured metadata for agent registries
    Typed schemas, compliance status and metadata-aware search make registry records easier to organise and discover.
  • One-response MCP tool lists
    AgentCore Gateway can return the complete tools list without pagination, though the feature is in limited preview.
  • Penetration tests in CI/CD
    SecurityAgent can run automated tests on code changes and gate deployment on findings.
  • Device posture and Cedar for Client VPN
    The SDK adds support for these access-assessment and authorisation capabilities.
  • Streaming configuration for Amazon Connect
    StartChatContact can now accept chat-streaming settings and return connection details in the same call.

Why it matters

These are not merely extra fields in a sprawling SDK update. Registry metadata can help teams describe and find the tools their agents can use, while the AgentCore change simplifies retrieving those tools. Together, they address some unglamorous but important plumbing around agent discovery and integration.

The CI/CD testing and VPN changes also give developers more ways to bring security checks and access policies into existing workflows. The release notes describe available SDK support, not evidence of how well those controls perform in deployment; still, they show where AWS is making its interfaces easier to build against.

Our read

This is a useful, multi-part release for teams building on AWS, particularly those wiring agents to registries and MCP tools. The best signal is the focus on discoverability and enforceable workflow controls, rather than another promise that agents will simply behave themselves. Developers should check the limited-preview status before relying on AgentCore’s full-list response in production.

What to watch

  • Whether AgentCore’s complete MCP tools list moves beyond limited preview.
  • How teams use registry schemas and compliance status in practice.
  • Whether AWS publishes further detail on SecurityAgent test coverage and results.
  • Which SDK and service changes follow for agent discovery and access control.

Discussion spark: For agent systems, which would make the bigger practical difference: better tool discovery through registry metadata, or security checks that can block a deployment?

Sources and evidence

not affiliated with or endorsed by Amazon Web Services (AWS)

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.