AWS SDK for .NET V3 Reaches End-of-Support
AWS SDK for .NET V3 reached end-of-support on June 1, 2026. There will be no further updates, including security fixes, for V3. AWS had entered V3 into maintenance mode in March 2026, giving teams three months to plan migrations, but the June 1 date is now firm: any V3 dependency on an internet-facing or regulated workload carries unpatched risk going forward.
Migration to AWS SDK for .NET V4 is the recommended path. V4 delivers performance improvements including reduced memory allocations in common data-transfer paths, updated async patterns that align with modern .NET idioms, and continued parity with new AWS service APIs. Teams using dependency injection wrappers such as AWSSDK.Extensions.NETCore.Setup will need to update package references and, in some cases, adapt to revised client-factory interfaces introduced in V4.
The practical blast radius depends on the .NET version in use. Teams on .NET 8+ (LTS) should encounter the cleanest migration path, as V4 was designed around that runtime. Teams still on .NET Framework face a harder decision, since V4 targets .NET 6+ and does not support the full .NET Framework surface. AWS documentation for V3-to-V4 migration is available on the AWS Developer Tools blog and covers the most common breaking changes by service category.
Read more — AWS Developer Tools Blog
Amazon Aurora MySQL Gains Natural-Language Control via Kiro Powers
Amazon Aurora MySQL now integrates with Kiro Powers, the natural-language AI layer that AWS introduced with the Kiro IDE assistant. Developers working inside Kiro can execute both data plane operations (queries, DML) and control plane tasks (Serverless scaling configuration, RDS-to-Aurora migration steps, replication setup) in conversational language, with Kiro translating intents to the appropriate AWS API calls and providing contextual guidance.
The integration extends Kiro's spec-driven development model into the database tier. A developer describing a migration scenario — for example, moving a multi-region read-replica setup from RDS MySQL to Aurora Serverless v2 — gets back a step-by-step action plan with the specific Aurora API calls, estimated capacity units, and caveats, rather than having to cross-reference documentation. Kiro Powers also surfaces dynamic scaling recommendations based on observed CloudWatch metrics, helping teams right-size Aurora Serverless capacity without manually reviewing charts.
For Java and Spring teams, the Aurora MySQL Kiro integration pairs naturally with Spring Data JPA's Aurora-compatible datasource configuration and the auto-scaling annotations available through Spring Cloud AWS. The natural-language control plane reduces the operational overhead of managing Aurora beyond the application layer, which has historically required a separate DevOps or DBA function for anything beyond basic connection string management.
Read more — AWS News Blog
Codex and GPT-5.5 Now Generally Available on Amazon Bedrock
OpenAI's Codex coding agent and GPT-5.5 are now generally available on Amazon Bedrock, expanding access beyond the limited preview that launched in May 2026. GPT-5.5 and GPT-5.4 run on Bedrock's next-generation inference engine and are available through the Responses API at pricing that matches OpenAI's direct rates with no additional Bedrock surcharge.
All inference routes through Amazon Bedrock's existing security perimeter: IAM policies, VPC isolation, PrivateLink, encryption at rest and in transit, and CloudTrail logging apply to OpenAI model calls just as they do to Amazon's own models. For enterprises with strict data-residency or audit requirements that have blocked direct OpenAI API usage, this route provides access to frontier models through infrastructure the organisation's security team has already approved.
Codex is accessible through the Codex app, CLI, and IDE integrations with inference routed through Bedrock, meaning the same IAM role-based access controls that govern a team's Lambda and S3 access can govern which developers and CI pipelines can invoke Codex. Teams already using Amazon Bedrock AgentCore or Bedrock Managed Agents can integrate Codex as a worker agent in multi-agent pipelines without adding a separate OpenAI API key to their secret management stack.
Read more — AWS News Blog
