awslabs/agent-plugins

AWS Architecture Diagram generates validated draw.io XML for AWS system and cloud infrastructure diagrams. Use it to create new diagrams or turn CloudFormation, CDK, or Terraform into visuals with official AWS4 icons.

AWS Lambda helps design, build, deploy, test, and debug serverless applications with AWS Lambda and related services. Use it for Lambda functions, event sources, API Gateway, EventBridge, Step Functions, and other event-driven architectures.

AWS Lambda Durable Functions builds resilient, long-running multi-step applications with state persistence, retries, waits, and callbacks. Use it for durable serverless workflows, saga patterns, and local testing with the Durable test runner.

AWS Lambda Managed Instances evaluates, configures, and migrates workloads to Lambda on customer-managed EC2 capacity. Use it for steady, high-volume workloads, scheduled scaling, or comparing LMI with standard Lambda and EC2.

AWS Lambda MicroVMs builds and runs Firecracker-isolated serverless environments with snapshot resume. Use it for sandboxed code execution, interactive notebooks, long-lived sessions, or other port-listening workloads.

AWS Serverless Deployment helps scaffold and deploy serverless apps on AWS with SAM or CDK. Use it for templates, constructs, deployment workflows, and serverless CI/CD; for Lambda runtime behavior, use the aws-lambda skill.

AWS Step Functions helps you build and transform state machine workflows in ASL using JSONata. Use it to define states, handle errors, integrate AWS services, and migrate from JSONPath.

AWS Transform migrates, modernizes, and upgrades codebases to AWS. It analyzes repos for tech debt, security issues, and modernization opportunities such as .NET, COBOL, VMware, SQL Server, Java, Python, Node.js, and AWS SDK upgrades.

Amazon API Gateway helps you build, manage, and troubleshoot REST, HTTP, and WebSocket APIs on AWS. Use it when working with custom domains, authorizers, usage plans, throttling, CORS, private APIs, or API Gateway IaC templates.

Amazon Location Service integrates AWS geospatial APIs for maps, geocoding, routing, places search, geofences, and tracking. Use it when building location-aware applications or adding map and location features to AWS apps.

Amplify Workflow builds and deploys full-stack web and mobile apps with AWS Amplify Gen2. Use it when setting up backend resources, connecting frontends, or deploying Amplify-based projects.

Aurora DSQL helps you build with Amazon Aurora DSQL by managing schemas, running queries, handling migrations, and diagnosing performance. Use it for IAM auth, multi-tenant patterns, bulk loads, and SQL compatibility checks.

Dataset Evaluation checks your dataset’s file type, schema, and format for SageMaker fine-tuning or evaluation. Use it before training or when you need to know whether your data is valid for a selected model and strategy.

Dataset Transformation generates code to convert datasets between ML training and evaluation schemas. Use it when a dataset needs to be reformatted for OpenAI, SageMaker, HuggingFace, Bedrock Nova, VERL, or custom JSONL outputs.

Deploy On AWS analyzes an application and chooses suitable AWS services, then estimates cost and generates infrastructure as code with security defaults. Use it when you want to host or run a codebase on AWS with minimal decisions.

Directory Management resolves a project name and creates a standard folder structure for plans, scripts, notebooks, manifests, and agent memory. Use it when starting a new project or resuming an existing one with a PLAN.md.

Document Service analyzes an existing codebase and produces structured technical documentation and architecture diagrams with file and line citations. It is used when you need to understand, onboard to, or document an inherited system, especially AWS workloads.

Elastic Beanstalk deploys web and worker applications to AWS with managed scaling, patching, health checks, and deployments. Use it when you want a standard application runtime without managing servers or container orchestration.

Finetuning generates code for SageMaker serverless training jobs to fine-tune a base model. Use it when starting training or when a plan reaches the finetuning step, including SFT, DPO, RLVR, and RLAIF workflows.

Finetuning Technique helps choose and validate a fine-tuning method such as SFT, DPO, RLVR, or RLAIF for a selected base model. Use it after model selection when you need to match a technique to a use case and supported recipes.

HyperPod Cluster Debugger diagnoses read-only HyperPod cluster failures for EKS or Slurm, including provisioning, access, CloudFormation, capacity, and node issues. It uses `--validate` for pre-flight checks before creation or remediation.

HyperPod Issue Report collects diagnostic logs and configuration from HyperPod EKS or Slurm clusters and stores them in S3. Use it to troubleshoot node failures, capture cluster state, or prepare AWS Support cases.

HyperPod NCCL Debugger diagnoses NCCL and related training-pod failures on SageMaker HyperPod GPU clusters. Use it for hangs, collective timeouts, EFA or libfabric errors, rendezvous failures, OOMs, DNS issues, and pod scheduling problems.

HyperPod Performance Debugger diagnoses uneven NCCL bandwidth across nodes and slow filesystem throughput on Amazon SageMaker HyperPod. It gathers read-only host signals and routes hardware, version, or NCCL issues to sibling skills.

Hyperpod Node Debugger diagnoses and helps remediate a single unhealthy HyperPod node in EKS or Slurm. Use it for node-level EFA, GPU, disk, memory, SSM, runtime, or lifecycle-script issues; it is read-only and not for cluster-wide provisioning.