← Back to all announcements
★★★☆☆ 09/07/2026

OAuth support for the AWS MCP Server

AI agents can now authenticate to AWS MCP Server via standard OAuth — no CLI or extra software needed, with full IAM governance intact.

View original announcement →

Visual Summary

graph TD A{{OAuth for AWS MCP Server}}:::announced B((AI Agents)):::external C(AWS Sign-In):::compute D([Interactive Auth]):::feature E([Headless Auth]):::feature F(IAM Policies):::compute G(AWS CloudTrail):::storage H([Token Management]):::feature I(AWS MCP Server):::compute B ==>|"requests access"| A A -->|"authenticates via"| C C -->|"browser flow"| D C -->|"programmatic flow"| E A -->|"issues tokens"| H A -->|"enforces"| F A -.->|"audits events"| G H -->|"authorizes"| I classDef announced fill:#ff9900,stroke:#ec7211,color:#fff,font-weight:bold classDef compute fill:#e3f2fd,stroke:#1565c0,color:#1565c0 classDef storage fill:#e8f5e9,stroke:#2e7d32,color:#2e7d32 classDef feature fill:#fff3e0,stroke:#e65100,color:#e65100 classDef external fill:#f5f5f5,stroke:#616161,color:#616161

What's New

AWS has introduced OAuth 2.0 support for the AWS MCP (Model Context Protocol) Server, enabling AI agents to authenticate directly using AWS Sign-In without requiring additional authentication software. Agents can now connect using industry-standard OAuth flows — either interactively via a browser or programmatically via headless (non-interactive) authorization. Existing AWS identities, IAM permissions, federation configurations, and governance controls all continue to apply seamlessly.

How It Works

  • OAuth Authorization Flow: An agent or application requests access to the AWS MCP Server; AWS Sign-In authenticates the user and issues short-lived OAuth access tokens and refresh tokens that authorize the agent to act on behalf of the authenticated identity.
  • Interactive Authorization: For human-in-the-loop scenarios, the agent opens a browser redirecting to the familiar AWS Sign-In page; the user authenticates (supporting IAM federation, IAM Identity Center, root, or IAM users) and approves the authorization request.
  • Non-Interactive (Headless) Authorization: Automated agents with existing AWS credentials can obtain OAuth tokens programmatically via a new API, without requiring a browser session.
  • IAM Integration: OAuth tokens do not grant additional permissions beyond what the underlying IAM identity already has; all requests are evaluated against existing IAM policies, SCPs, RCPs, and permission boundaries.
  • New Governance Capabilities: Administrators gain new global IAM condition keys for OAuth, token introspection and revocation APIs, dynamic client registration for approved OAuth applications, and new CloudTrail audit events covering OAuth authorization and token lifecycle.
  • Required Permissions: The managed policy AWSMCPSignInOAuthAccessPolicy must be attached to the IAM role to enable OAuth-based sign-in for the MCP Server.
  • Endpoint Configuration: Agents are pointed to the AWS MCP Server endpoint (e.g., https://aws-mcp.us-east-1.api.aws/mcp) via HTTP transport, compatible with MCP-supporting agents like Claude Code, Kiro, Codex, and Gemini.

Why It's Important

  • Lowers the barrier to agentic AWS access: Developers can connect AI agents to AWS services without installing the AWS CLI, configuring local credentials, or deploying additional authentication middleware — dramatically simplifying onboarding.
  • Reuses existing identity infrastructure: Organizations don't need to build or maintain separate identity systems for agents; existing IAM roles, federated identities, and SSO configurations work out of the box.
  • Enables web-based agent clients: OAuth support means browser-based AI clients (e.g., Claude.ai, ChatGPT.com) can now connect to the AWS MCP Server, a scenario previously impossible with SigV4-only authentication.
  • Strengthens security posture: Short-lived tokens, token revocation, introspection APIs, and CloudTrail audit events give security teams full visibility and control over agent access — critical for enterprise governance.
  • Standardizes agent authentication: Using industry-standard OAuth 2.0 means tooling, libraries, and practices from the broader ecosystem apply directly, reducing custom integration work.
  • Supports the growing MCP ecosystem: As MCP becomes a de facto standard for agent-tool connectivity, native OAuth support positions AWS as a first-class MCP server compatible with any compliant agent framework.

How It's Different

  • vs. SigV4 (existing method): SigV4 requires the AWS CLI, local credential configuration, and the MCP Proxy for AWS; OAuth requires none of these, making it significantly simpler for new users and web-based clients.
  • vs. custom agent auth solutions: Rather than building bespoke authentication layers for agents, OAuth leverages the same AWS Sign-In infrastructure used by the AWS Console and CLI — no new identity plane to manage.
  • vs. static long-lived credentials: OAuth issues short-lived tokens with built-in expiry and revocation capabilities, reducing the risk surface compared to long-lived access keys.
  • New governance primitives: The addition of OAuth-specific global condition keys, dynamic client registration, and token introspection APIs are net-new capabilities not previously available in the AWS IAM/Sign-In ecosystem.
  • Session reuse: If a user already has an active AWS Sign-In session (e.g., from a prior Console login), the OAuth flow can reuse that session — eliminating redundant authentication steps.
  • Dual-mode support: Unlike solutions that force a single auth model, AWS MCP Server supports both OAuth and SigV4, letting teams choose the right method per use case without forklift changes.

When to Prefer It

  • New AWS users or simple single-account setups: OAuth is the recommended starting point — no CLI installation or credential configuration required.
  • Web-based agent clients: If your agent runs in a browser-based environment (Claude.ai, ChatGPT.com, or similar), OAuth is the only viable option since these clients cannot run a local MCP proxy.
  • Interactive, human-supervised agent workflows: When a human is present to approve authorization in a browser, the interactive OAuth flow provides a familiar and auditable sign-in experience.
  • Automated pipelines with existing AWS credentials: Use non-interactive (headless) OAuth when running agents in CI/CD or server-side environments that already have AWS credentials but need OAuth tokens for MCP access.
  • Organizations prioritizing audit and governance: When detailed CloudTrail logging of agent token issuance, introspection, and revocation events is required for compliance.
  • Avoid OAuth (prefer SigV4) when: You need to switch between multiple AWS accounts in a single session, require read-only mode to hide write-capable tools, need to set a default region per session, or your organization restricts the signin:AuthorizeOAuth2Access and signin:CreateOAuth2Token permissions.

Availability

  • General Availability: This feature is generally available as of July 9, 2026.
  • Supported Regions: The AWS MCP Server endpoint is currently available in us-east-1 based on the documented endpoint (https://aws-mcp.us-east-1.api.aws/mcp); additional regional endpoints may be available — consult the official documentation for the current list.
  • Supported Identity Types: IAM federation, AWS IAM Identity Center, root users, and IAM users are all supported for OAuth-based sign-in.
  • Pricing: No separate pricing is announced for OAuth support itself; standard AWS Sign-In and IAM usage applies. Costs associated with underlying AWS services accessed via the MCP Server follow their respective pricing models.
  • Limitations: Multi-account switching within a single OAuth session is not supported (use SigV4 for that scenario); read-only mode (hiding write-capable tools) is also a SigV4-only capability; some organizations with restrictive IAM policies blocking signin:AuthorizeOAuth2Access or signin:CreateOAuth2Token will need policy updates or must use SigV4.
  • Required Managed Policy: AWSMCPSignInOAuthAccessPolicy must be attached to the IAM role used for OAuth-based MCP Server access.

Tags

Servicesother-aws
Typenew-featuresecurity
Conceptsagentic-ai
Use Casesenterprisedeveloper-tools
GeographyGlobal

Related Resources

AI Radar AWS

AWS AI/ML news — curated, researched, explained

An automated intelligence platform that curates, researches, and analyzes AWS AI/ML/GenAI announcements daily. Every report is backed by real research — the system reads linked blog posts and documentation to provide accurate, in-depth analysis.

How Each Report Is Generated

  1. Collection — Daily monitoring of the AWS "What's New" RSS feed
  2. Filtering — AI-powered relevance detection for AI/ML/GenAI topics
  3. Taxonomy Tagging — LLM-based classification across 6 dimensions
  4. Importance Scoring — Point-based system with tag bonuses (1-5 stars)
  5. Research Phase — Follows links to blog posts and documentation
  6. Report Generation — Claude Sonnet produces structured 6-section analysis
  7. Visual Summary — Claude Opus generates Mermaid diagrams for key items
  8. Publishing — Static website rebuilt and deployed via CloudFront

Features

  • Faceted filtering by service, type, concept, and more
  • Multi-dimensional taxonomy with 80+ tags across 6 dimensions
  • Geographic availability badges (Global, APJ, EMEA, AMER) with filtering
  • Timeline visualization of announcement volume
  • PDF export for offline reading
  • Mermaid visual summaries for key announcements
  • Daily automated updates — no manual curation
What makes this different: Each report involves a dedicated research phase where the system reads linked blog posts and AWS documentation pages. This produces analysis that goes beyond the original announcement text.

Technology

Built with Python, AWS Lambda, Amazon Bedrock (Claude Sonnet 4.6, Opus 4.6, Haiku 4.5), S3, CloudFront, WAF, EventBridge, and CDK.

Open Source

This project is open source. Fork it, customize it for your needs, and deploy your own instance.
📦 github.com/bbonik/ai-radar-aws

How Importance Scoring Works

Each announcement receives a point score based on multiple factors. The total score maps to a 1-5 star rating:

1★ < 2 pts 2★ ≥ 2 pts 3★ ≥ 3.5 pts 4★ ≥ 5 pts 5★ ≥ 6.5 pts

Point Breakdown

FactorPointsWhen
Core AI service (Bedrock, AgentCore, SageMaker AI)+4Service named in title
Key AI service (SageMaker, Kiro, QuickSight)+2Service named in title
Other AI-related service+1Default
Blog post link+3Link to aws.amazon.com/blogs/
GitHub samples link+2Link to github.com/aws*
Documentation link+1Link to docs.aws.amazon.com/
New model+1.5Tagged as "new-model"
New service+1Tagged as "new-service"
New feature+0.5Tagged as "new-feature"
Anthropic / OpenAI provider+2Provider explicitly mentioned
Instance / notebook announcement-2Hardware/capacity, not feature
Performance / pricing / security-0.5Incremental updates
Region expansion to APJ+1Expands to Asia Pacific
Region expansion (non-APJ only)-1.5Only expands to other regions

Geographic Relevance Badges

Each announcement card shows a small badge indicating whether the feature is available in your region:

🌐 Global Available in all regions
🌏 APJ Asia Pacific
🌍 EMEA Europe / Middle East / Africa
🌎 AMER Americas (US, Canada, South America)
No badge Geography unknown
How geography is detected: The system detects ALL geographies mentioned in each announcement. If the text mentions specific regions (Tokyo, Frankfurt, Oregon, etc.), the corresponding geography badges are shown. If it says "all regions" or is a new feature with no region specified, it gets the Global badge. Geography is also filterable — click a geo chip to see only announcements available in that region.