News For Dev-ops

Suppose you have some C++/WinRT code that receives a delegate from an outside source, and you might invoke that delegate from a potentially different COM context. However, the original delegate may not be agile. How can you make an agile version of that delegate? The easy way is to wrap the delegate in an agile_ ref , and then resolve the agile_ ref back to a delegate when you want to invoke it. template<typename Delegate> Delegate make_agile_delegate(Delegate const& d) { return [agile = winrt::agile_ref(d)](auto&&...args) { return agile.get()(std::forward<decltype(args)>(args)...); }; } But if it were that easy, why would we call this article "part 1"? More in part 2. Post Updated on July 20, 2026 at 03:00PM Thanks for reading from devamazonaws.blogspot.com

Amazon CloudWatch announces the launch of coding agent insights, giving engineering leaders visibility into how AI coding tools are driving value across their organization. Coding Agent Insights integrates with Claude apps gateway for AWS to collect telemetry from Claude Code without additional instrumentation. Other supported coding agents include Codex and GitHub Copilot.  As organizations scale AI coding agent adoption, they need to understand return on investment. Coding agent insights is built on OpenTelemetry metrics emitted by your coding agents and presents them alongside your existing CloudWatch operational data. This helps you answer questions like which teams would benefit from expanded access, where are agents accelerating delivery, and how can you right-size token budgets across departments. You can track spend trends, set proactive token billing alerts, correlate agent adoption with improvements in commit throughput and pull request velocity, or identify the models...

If you use the Azure Cosmos DB vNext emulator, you probably know the local development loop: start the emulator, connect to it, create some resources, load test data, run queries, and inspect the results. Each step is straightforward, but together they add setup work before you can test the application you are actually building. How agents work with the emulator   The emulator includes the Azure Cosmos DB Shell , an open-source CLI for working with databases, containers, and items. It runs inside the emulator container and handles the local endpoint and well-known key, giving developers a direct, scriptable way to work with the emulator. CLIs are well suited to agent workflows because they expose operations as explicit commands and return results the agent can inspect. An agent can discover commands through built-in help, run them non-interactively, respond to errors, and use the output to choose its next action. These characteristics make the CLI a practical interface for age...

Amazon SageMaker HyperPod now supports network topology configuration at the partition level for Slurm orchestrated clusters. A single cluster can now run tree topology in one partition and block topology in another, with each partition using the topology best suited to its instance types. This improves distributed training performance by keeping job placement aligned with the interconnect characteristics of each instance type, so GPU-to-GPU communication is faster, NCCL collective operations are more efficient, and training throughput improves. HyperPod determines the topology for each partition based on the instance types of its compute instance groups. Partitions with Amazon EC2 UltraServer instance types such as ml.p6e-gb200.36xlarge use block topology, and those with hierarchical-interconnect instance types such as ml.p5.48xlarge, ml.p5e.48xlarge, and ml.p5en.48xlarge use tree topology, while partitions with instance types that don't provide network topology information rema...

Amazon GameLift Streams now supports assigning an IAM role to a stream session, enabling your application to securely access resources in your AWS account, such as Amazon S3 buckets and DynamoDB tables. With this launch, you can pass a RoleArn parameter when starting a stream session, and your application automatically receives short-lived, auto-refreshing AWS credentials through the standard AWS SDK credential resolution chain — no application code changes required. Previously, customers who needed their streamed applications to access AWS services had to embed long-lived access keys in application bundles or pass them as environment variables, creating security and operational challenges. Now, Amazon GameLift Streams handles credential vending and automatic refresh using the same container credential provider mechanism trusted by Amazon ECS task roles and Amazon EKS Pod Identity. Role misconfigurations are validated at session start, surfacing clear errors immediately rather than d...

You shipped a skill that calls an API. The agent uses it, user get results. But how do you know the results are good ? How do you know your last change didn't quietly break the happy path, or that switching models won't regress the scenarios you already got right? You need to evaluate. Run your skill through a set of scenarios, score the outputs, compare across iterations. Without that, you're shipping blind and hoping for the best. The problem nobody warns you about The moment you start evaluating a skill that calls an API, things get complicated. If that API belongs to an external service, every eval run costs money. Say you're running 50 scenarios across 3 models with 5 repetitions per scenario. That's at least 750 API calls per session. Multiply by every iteration as you tune prompts and try different models. The meter is running, and you're paying for test traffic that produces no user value. But even when the API is yours and costs nothing to c...