AWS bedrock-agentcore documentation change
Summary
Updated documentation with branding changes (AgentCore Runtime to Amazon Bedrock AgentCore Runtime), added links to related resources, expanded authentication/authorization details, and added protocol implementation guides.
Security assessment
The change adds explicit documentation about inbound authentication (IAM/OAuth 2.0) and outbound authentication (OAuth/API keys) security features, including credential management through AgentCore Identity. This provides new security documentation but doesn't indicate a response to a specific vulnerability.
Diff
diff --git a/bedrock-agentcore/latest/devguide/runtime-how-it-works.md b/bedrock-agentcore/latest/devguide/runtime-how-it-works.md index 303a7cd55..fb9d8b76c 100644 --- a//bedrock-agentcore/latest/devguide/runtime-how-it-works.md +++ b//bedrock-agentcore/latest/devguide/runtime-how-it-works.md @@ -9 +9 @@ Key componentsAuthentication and securityAdditional featuresImplementation overv -The AgentCore Runtime handles scaling, session management, security isolation, and infrastructure management, allowing you to focus on building intelligent agent experiences rather than operational complexity. By leveraging the features and capabilities described here, you can build, deploy, and manage sophisticated AI agents that deliver value to your users while helping to maintain enterprise-grade security and reliability. +The Amazon Bedrock AgentCore Runtime handles scaling, session management, security isolation, and infrastructure management, allowing you to focus on building intelligent agent experiences rather than operational complexity. By leveraging the features and capabilities described here, you can build, deploy, and manage sophisticated AI agents that deliver value to your users while helping to maintain enterprise-grade security and reliability. @@ -21,2 +20,0 @@ The AgentCore Runtime handles scaling, session management, security isolation, a - * [Understanding the AgentCore Runtime service contract](./runtime-service-contract.html) - @@ -32 +30 @@ An AgentCore Runtime is the foundational component that hosts your AI agent or t -You can build and deploy agents to AgentCore Runtime using the AgentCore Python SDK or directly through AWS SDKs. With the AgentCore Python SDK, you can define your agent using popular frameworks like LangGraph, CrewAI, or Strands Agents. The SDK handles infrastructure complexities, allowing you to focus on the agent's logic and capabilities. +You can build and deploy agents to AgentCore Runtime using the [AgentCore Runtime starter toolkit](./develop-agents.html#bedrock-agentcore-configure-deploy-starter-toolkit), the [AgentCore Python SDK](./develop-agents.html#develop-agents-bedroock-agentcore-sdk) or directly through [AWS SDKs](./develop-agents.html#develop-agents-bedrock-agentcore-aws-sdk). With the AgentCore Python SDK, you can define your agent using [popular frameworks](./using-any-agent-framework.html) like LangGraph, CrewAI, or Strands Agents. The SDK handles infrastructure complexities, allowing you to focus on the agent's logic and capabilities. @@ -45 +43 @@ Each AgentCore Runtime: -Each AgentCore Runtime maintains immutable versions that capture a complete snapshot of the configuration at a specific point in time: +Each AgentCore Runtime maintains immutable [versions](./agent-runtime-versioning.html) that capture a complete snapshot of the configuration at a specific point in time: @@ -60 +58 @@ This versioning system provides reliable deployment history and rollback capabil -Endpoints provide addressable access points to specific versions of your AgentCore Runtime. Each endpoint: +[Endpoints](./agent-runtime-versioning.html) provide addressable access points to specific versions of your AgentCore Runtime. Each endpoint: @@ -103 +101 @@ You can update endpoints without downtime, allowing for seamless version transit -Sessions represent individual interaction contexts between users and your AgentCore Runtime. Each session: +[Sessions](./runtime-sessions.html) represent individual interaction contexts between users and your AgentCore Runtime. Each session: @@ -143,0 +142,2 @@ This complete isolation between sessions is crucial for enterprise security, par +Inbound authentication controls who can access your agents through AWS Identity and Access Management or OAuth 2.0, validating bearer tokens from identity providers before allowing requests to proceed. Outbound authentication enables your agents to securely access third-party services using OAuth or API keys, with AgentCore Identity managing credentials in either user-delegated or autonomous modes. For more information, see [Authenticate and authorize with Inbound Auth and Outbound Auth](./runtime-oauth.html). + @@ -209 +209 @@ Outbound Auth, powered by Amazon Bedrock AgentCore Identity, lets your agents ho - * Enterprise systems such as Slack, Zoom, and GitHub) + * Enterprise systems such as Slack, Zoom, and GitHub @@ -234,0 +235,2 @@ AgentCore Runtime supports long-running workloads through: +For more information, see [Handle asynchronous and long running agents with Amazon Bedrock AgentCore Runtime](./runtime-long-run.html). + @@ -237 +239 @@ AgentCore Runtime supports long-running workloads through: -Agents can stream partial results as they become available rather than waiting for complete processing. This lets you provide a more responsive user experience, especially for operations that generate large amounts of content or take significant time to complete. +Agents can stream partial results as they become available rather than waiting for complete processing. This lets you provide a more responsive user experience, especially for operations that generate large amounts of content or take significant time to complete. For more information, see [Stream agent responses](./response-streaming.html). @@ -241 +243,3 @@ Agents can stream partial results as they become available rather than waiting f -Runtime supports multiple communication protocols: +AgentCore Runtime supports multiple communication protocols: + + * [HTTP](./runtime-http-protocol-contract.html): Direct REST API endpoints for traditional request/response patterns. For more information, see [Get started with the Amazon Bedrock AgentCore starter toolkit](./runtime-get-started-toolkit.html). @@ -243 +247 @@ Runtime supports multiple communication protocols: - * HTTP for simple request/response patterns + * [MCP](./runtime-mcp-protocol-contract.html): Model Context Protocol for tools and agent servers. For more information, see [Deploy MCP servers in AgentCore Runtime](./runtime-mcp.html). @@ -245 +249 @@ Runtime supports multiple communication protocols: - * Model Context Protocol (MCP) for standardized agent-tool interactions + * [A2A](./runtime-a2a-protocol-contract.html): Agent-to-Agent protocol for multi-agent communication and discovery. For more information, see [Deploy A2A servers in AgentCore Runtime](./runtime-a2a.html). @@ -252 +256 @@ Runtime supports multiple communication protocols: -Here's how to get started with the AgentCore Runtime: +Here's how to get started with the AgentCore Runtime. For the complete example, see [Get started with the Amazon Bedrock AgentCore starter toolkit](./runtime-get-started-toolkit.html). @@ -308 +312 @@ AgentCore Runtime: Host agent or tools -Understanding the AgentCore Runtime service contract +Understand the AgentCore Runtime service contract