Understand this story
On November 18, 2026, Nithyashree Alwarsamy published a post on the AWS Machine Learning blog introducing three OAuth configurations tested on Amazon Bedrock AgentCore Identity, used to enable AgentCore Gateway to securely call Snowflake-hosted MCP servers. The three configurations are: Snowflake OAuth (3LO, no external IdP required), third-party IdP three-legged OAuth (enterprise SSO), and third-party IdP two-legged OAuth (shared service user). The post explains that AgentCore Identity handles token custody, refresh, and binding of users to tokens, and supports on-behalf-of (TOKEN_EXCHANGE) authorization, providing per-user identity for unattended workloads. Testing used Okta and Microsoft Entra ID, and the first two modes can be extended to other OAuth-protected MCP servers. Currently, this post is the only relevant report, with no subsequent developments yet.
AI-generated from public reports
Latest development · Report
Securely connecting to Snowflake Cortex AI through Amazon Bedrock AgentCore…Developments
1 confirmed developments
AWS blog introduces OAuth configuration for AgentCore integration with Snowflake Cortex AI
Source:AWS Machine Learning
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 18
AWS Machine LearningFirst-party
Securely connecting to Snowflake Cortex AI through Amazon Bedrock AgentCore…
Amazon Bedrock AgentCore Identity supports three OAuth configurations that let AgentCore Gateway securely call Snowflake-hosted MCP servers: Snowflake OAuth (3LO, no external IdP required), third-party IdP three-legged OAuth (enterprise SSO), and third-party IdP two-legged OAuth (shared service user). AgentCore Identity handles token custody, refresh, and user-to-token binding, and supports on-behalf-of (TOKEN_EXCHANGE) authorization to provide per-user identity for unattended workloads. Testing used Okta and Microsoft Entra ID, and the first two modes can be extended to other OAuth-protected MCP servers.
AI Trends