AWS tests three OAuth configurations for AgentCore integration with Snowflake
Understand this story
AWS published a tutorial explaining that it tested three OAuth configurations on Amazon Bedrock AgentCore Identity for accessing Snowflake-managed MCP servers for Snowflake Cortex AI: Snowflake OAuth (per-user authorization without an external identity provider), third-party identity provider three-legged OAuth (enterprise single sign-on), and third-party identity provider two-legged OAuth (shared service user). The tests used Okta and Microsoft Entra ID. AWS says the latter two external identity provider modes follow standard OAuth flows and can be extended to other OAuth-protected MCP servers, but token claims, scopes, and audience settings need to be validated for the specific SaaS provider. Although user-identity-based OBO is already generally available in AgentCore Identity, it is not within the scope of testing for these three configurations. The reference implementation for the demo exposes SYSTEM_EXECUTE_SQL and Cortex Agent on the same MCP server.
AI overview based on source reports · Updated 4 hr ago
Coverage timeline
Follow the coverage to explore different perspectives.
- AWS Machine Learning BlogSecure OAuth patterns for Snowflake Cortex AI through Amazon Bedrock AgentCore
The post describes three OAuth configurations tested on Amazon Bedrock AgentCore Identity for outbound authentication from Amazon Bedrock AgentCore Gateway to the Snowflake-managed MCP server: Snowflake OAuth (per-user, no external IdP), three-legged OAuth with a third-party IdP via corporate SSO, and two-legged OAuth with a shared service user. It details each pattern's flow, Snowflake security integration settings, token refresh behavior, and trade-offs, noting that only the Snowflake OAuth pattern is Snowflake-specific. It also mentions AgentCore Identity's on-behalf-of (TOKEN_EXCHANGE) grant for per-user identity in unattended workloads.
Activity for this story
- Current comparable activity
- 9
- Peak comparable activity
- 10Nov 18 05:00
- Change in the past 24 hours
- –
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.