DevOps & Cloud Skills

>- Guides agents to interactively discover customer requirements for live, bidirectional multi-agent AI systems that process continuous streams of multimodal data for real-time technical guidance and safety monitoring. Generates a custom Google Cloud solution that uses opinionated best practices and architecture guidance. Use when users need agentic assistance to design and create a multi-product solution in the cloud for live bidirectional multimodal streaming workloads. Don't use for simple text-based chat applications or workloads without real-time streaming requirements.

>- Designs a tailored multi-product agentic data science architecture on Google Cloud that incorporates opinionated best practices. Use when architecting multi-product solutions for agent-based data analytics or ML workloads. Don't use for simple queries, non-agentic pipelines, general cloud reviews, or writing agent code.

>- Guides a developer's first steps on Google Cloud, covering account creation, billing setup, project management, and deploying a first resource. Use when a new developer wants to initialize their first Google Cloud project, configure billing, and verify deployment. Don't use for enterprise organization setup (use Google Cloud Setup guided flow for that instead). Don't use for complex multi-project architectures.

>- Designs, builds, and deploys AI agents or multi-agent systems on Google Cloud. Provides an interactive workflow to gather requirements, recommend a tailored architecture, and generate deployment instructions. Use when designing or implementing agentic systems on Google Cloud. Don't use for general Google Cloud solution architecture (use google-cloud-solution-architecture instead) or for narrow tasks targeting a single product without agent context.

>- Guides the migration of existing AI workloads (Cloud Run, Gemini API, Gemini Enterprise Agent Platform) to self-hosted GKE inference using gcloud and kubectl. Use when the user has an existing AI inference workload (on Cloud Run, the Gemini API, Gemini Enterprise Agent Platform, or a custom VM) and wants to move it to self-hosted inference on GKE, or asks follow-up questions during such a migration (hardware sizing, model staging, manifest generation, validation, traffic cutover). DO NOT use for brand new GKE inference deployments with no existing workload to migrate (use gke-inference instead). DO NOT use if the user intends to automate the migration via the Gemini Cloud Assist MCP server.

>- Guides agents to interactively discover customer requirements for a Secure n-tier serverless web application and generate a tailored cloud multi-product solution that incorporates opinionated best practices and architecture guidance. Use when users need agentic assistance with designing and creating a multi-product solution in the cloud for Secure n-tier serverless web application. Don't use when designing VM or GKE-based architectures or when not using Google Cloud.

Generates cost optimization guidance for Google Cloud workloads based on the Google Cloud Well-Architected Framework (WAF). Use this skill to evaluate a workload, identify cost requirements and constraints, and provide actionable recommendations for build, deploy, and manage the workload cost-efficiently in Google Cloud.

>- Generates operations-focused guidance for Google Cloud workloads based on the design principles and recommendations in the Operational Excellence pillar of the Google Cloud Well-Architected Framework (WAF). Use this skill to evaluate a workload, identify operational requirements, and provide actionable recommendations for deployment, monitoring, and incident management.

>- Stores, retrieves, and manages data as objects in Cloud Storage (Google Cloud Storage, or GCS) buckets. Use when you need to interact with Cloud Storage — create or configure buckets, upload, download, stream, or transfer data, organize objects with folders, generate signed URLs, control access (IAM, ACLs, public access prevention), set storage classes and tiering (Standard, Nearline, Coldline, Archive), manage cost and lifecycle, protect data (versioning, encryption/CMEK, retention and Bucket Lock, object holds, soft delete), host static websites, trigger Pub/Sub notifications on object changes, mount buckets as a file system (gcsfuse), or optimize storage performance at any scale. Covers the gcloud storage / gsutil CLI, JSON and XML APIs, client libraries, Terraform, and Cloud Storage MCP servers. Don't use for block storage (Persistent Disk), data warehousing/analytics (BigQuery), or databases (Cloud SQL, Spanner, Bigtable, Firestore).

>- Generates sustainability-focused guidance for Google Cloud workloads based on the design principles and recommendations in the Google Cloud Well-Architected Framework (WAF). Use this skill to evaluate a workload, identify environmental impact requirements, and provide actionable recommendations to build, deploy, and manage the workload sustainably in Google Cloud.

> Enable and interpret TensorRT-LLM AutoDeploy FX graph text dumps via AD_DUMP_GRAPHS_DIR. Use when you need before/after graphs per transform, to locate subgraphs, or to confirm a rewrite ran. Paths and behavior are grounded in tensorrt_llm/_torch/auto_deploy (GraphWriter, BaseTransform). Complements ad-add-fusion-transformation.

Compile TensorRT-LLM on a compute node inside a Docker container. Use this when already on a compute node with GPUs visible.

Compile TensorRT-LLM on a SLURM cluster. Covers submitting a batch job with a container image, monitoring the job, and verifying the build. Use when the user wants to compile TRT-LLM remotely via SLURM rather than on a local compute node.

Workflow 1.5: Bridge between idea discovery and auto review. Reads EXPERIMENT_PLAN.md, implements experiment code, deploys to GPU, collects initial results. Use when user says \"实现实验\", \"implement experiments\", \"bridge\", \"从计划到跑实验\", \"deploy the plan\", or has an experiment plan ready to execute.

Full patent drafting pipeline from invention description to jurisdiction-formatted filing documents. Supports CN (CNIPA), US (USPTO), EP (EPO). Supports invention patents and utility models. Use when user says \"写专利\", \"patent pipeline\", \"专利申请\", \"draft patent\", \"写权利要求书\", or wants to draft a complete patent application.

Workflow 5: orchestrate a text-only resubmit of a polished paper to a different venue under hard constraints (no new experiments, no bib edits, no framework changes, never overwrite prior submissions). Use when user says \"resubmit pipeline\", \"重投流程\", \"port paper to <new venue>\", \"resubmit to <venue>\", \"tighten paper for resubmission\", or has a rejected/withdrawn paper to move to a different top venue under tight time budget.

Deploy and run ML experiments on local, remote, Vast.ai, or Modal serverless GPU. Use when user says "run experiment", "deploy to server", "跑实验", or needs to launch training jobs.

Run GPU workloads on Modal — training, fine-tuning, inference, batch processing. Zero-config serverless: no SSH, no Docker, auto scale-to-zero. Use when user says \"modal run\", \"modal training\", \"modal inference\", \"deploy to modal\", \"need a GPU\", \"run on modal\", \"serverless GPU\", or needs remote GPU compute.

Workflow 1.5: Bridge between idea discovery and auto review. Reads EXPERIMENT_PLAN.md, implements experiment code, deploys to GPU, collects initial results. Use when user says \"实现实验\", \"implement experiments\", \"bridge\", \"从计划到跑实验\", \"deploy the plan\", or has an experiment plan ready to execute.

Full patent drafting pipeline from invention description to jurisdiction-formatted filing documents. Supports CN (CNIPA), US (USPTO), EP (EPO). Supports invention patents and utility models. Use when user says \"写专利\", \"patent pipeline\", \"专利申请\", \"draft patent\", \"写权利要求书\", or wants to draft a complete patent application.

Workflow 5: orchestrate a text-only resubmit of a polished paper to a different venue under hard constraints (no new experiments, no bib edits, no framework changes, never overwrite prior submissions). Use when user says \"resubmit pipeline\", \"重投流程\", \"port paper to <new venue>\", \"resubmit to <venue>\", \"tighten paper for resubmission\", or has a rejected/withdrawn paper to move to a different top venue under tight time budget.

Deploy and run ML experiments on local or remote GPU servers. Use when user says \"run experiment\", \"deploy to server\", \"\u8dd1\u5b9e\u9a8c\", or needs to launch training jobs.

Run GPU workloads on Modal — training, fine-tuning, inference, batch processing. Zero-config serverless: no SSH, no Docker, auto scale-to-zero. Use when user says \"modal run\", \"modal training\", \"modal inference\", \"deploy to modal\", \"need a GPU\", \"run on modal\", \"serverless GPU\", or needs remote GPU compute.

Resolve missing setup caused by a hardcoded Foundry project endpoint or model in a sample. Use when a sample fails because it uses a placeholder/hardcoded project_endpoint (for example "https://your-project.services.ai.azure.com") or a hardcoded model instead of reading them from the environment.

> Step-by-step process for validating a Python Foundry hosted agent sample (under python/samples/04-hosting/foundry-hosted-agents/) end to end — running it locally (native runtime and `azd ai agent run`) and after deploying it to an Azure AI Foundry project with `azd`. Use this when asked to validate a hosted agent sample.

Use this skill to interact with the OSS-Fuzz infrastructure.

Plans and executes safe Convex schema and data migrations using the widen-migrate-narrow workflow and the @convex-dev/migrations component. Use this skill when a deployment fails schema validation, existing documents need backfilling, fields need adding or removing or changing type, tables need splitting or merging, or a zero-downtime migration strategy is needed. Also use when the user mentions breaking schema changes, multi-deploy rollouts, or data transformations on existing Convex tables.

Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling.

GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.

Half-Quadratic Quantization for LLMs without calibration data. Use when quantizing models to 4/3/2-bit precision without needing calibration datasets, for fast quantization workflows, or when deploying with vLLM or HuggingFace Transformers.

Evaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution. Use when needing scalable evaluation on local Docker, Slurm HPC, or cloud platforms. NVIDIA's enterprise-grade platform with container-first architecture for reproducible benchmarking.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Track ML experiments, manage model registry with versioning, deploy models to production, and reproduce experiments with MLflow - framework-agnostic ML lifecycle platform

Autonomous AI agent platform for building and deploying continuous agents. Use when creating visual workflow agents, deploying persistent autonomous agents, or building complex multi-step AI automation systems.

LLM observability platform for tracing, evaluation, and monitoring. Use when debugging LLM applications, evaluating model outputs against datasets, monitoring production systems, or building systematic testing pipelines for AI applications.

Open-source AI observability platform for LLM tracing, evaluation, and monitoring. Use when debugging LLM applications with detailed traces, running evaluations on datasets, or monitoring production AI systems with real-time insights.

Merge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or experimenting rapidly with model variants. Covers SLERP, TIES-Merging, DARE, Task Arithmetic, linear merging, and production deployment strategies.

Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with limited compute. Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.

Deploy a Microsoft Agent Framework (MAF) workflow as a managed online endpoint to an Azure ML workspace or an Azure AI Foundry hub-based project. Wraps any workflow into an init()/run() scoring script, creates conda environment, endpoint and deployment YAMLs, deploy script, and assigns RBAC. Supports managed identity auth and Application Insights tracing. WHEN: deploy MAF workflow, deploy agent-framework workflow, create online endpoint for MAF, deploy workflow to AML, deploy workflow to Foundry project, managed online endpoint for agent workflow, wrap workflow in scoring script, deploy agent as endpoint, realtime endpoint in Foundry project.

Convert an existing Prompt Flow Parallel Run Step (PRS) pipeline submission into an Azure ML PRS pipeline that runs a Microsoft Agent Framework (MAF) workflow. Wraps the MAF workflow into a PRS init()/run() entry script, generates the parallel component YAML and conda environment, and rewrites the pipeline submission script. Replaces what `load_component(flow.dag.yaml)` did automatically for Prompt Flow \u2014 produces the hand-built equivalent so that downstream pipeline code (`flow_node = flow_component(...)`, `flow_node.outputs.flow_outputs`, `flow_node.outputs.debug_info`, `flow_node.mini_batch_size`, scheduler, batch endpoint) stays unchanged. WHEN: convert promptflow PRS to MAF PRS, migrate PRS pipeline to agent framework, wrap MAF workflow as parallel component, bulk run MAF workflow, run agent framework as parallel run step, batch run MAF workflow on AML, submit MAF workflow as pipeline component, replace flow.dag.yaml with MAF workflow in pipeline, load_component equivalent for MAF workflow, MAF version of flow_component, load MAF workflow as component, wrap MAF workflow as flow component, MAF flow component, replace flow_node in pipeline with MAF workflow, keep flow_outputs and debug_info ports with MAF, MAF parallel component with connections={}, run MAF workflow as flow_node in AML pipeline, load_component('workflow.py') doesn't work. DO NOT USE FOR: converting the flow itself (use promptflow-to-maf), deploying as online endpoint (use maf-online-endpoint), enabling tracing only (use maf-tracing).

Enable tracing and logging for Microsoft Agent Framework (MAF) workflows. Configures OpenTelemetry export to Azure Application Insights and/or a generic OTLP endpoint using environment variables. Adds required packages to requirements.txt. WHEN: enable tracing, add tracing, enable logging, add logging, configure telemetry, Application Insights for MAF, OTLP export, observe workflow, monitor agent workflow, trace agent framework, instrument MAF, add observability, trace workflow executions, debug workflow.

Fix Terraform provider end user documentation issues detected by swissshepherd (ss). Removes an ignored target from the config, runs ss, validates findings, fixes the documentation, and commits.

Create a SageMaker endpoint (real-time, real-time scale-to-zero, or async) with autoscaling, CloudWatch alarms, and tagging enabled by default. Use this skill whenever about to create a SageMaker endpoint, write deployment code that calls `create_endpoint`, or finalize a deployment after the image URI and IAM role are known. Provides deploy.py for real-time endpoints, deploy_ic.py for real-time endpoints that scale to zero instances via inference components, and deploy_async.py for async endpoints (also scale-to-zero). This is the last step in the SageMaker deployment workflow. Never generate a bare `create_endpoint` call without these defaults — endpoints without autoscaling or alarms are demos, not deployments.

Plan and coordinate the deployment of a model to Amazon SageMaker AI. Use this skill whenever the user wants to deploy, host, serve, or expose a model on SageMaker or AWS — including phrases like "deploy a model", "host this LLM on AWS", "serve this embedding model", "deploy a reranker", "deploy a text-to-image / diffusion model", "host this for async inference", "create an endpoint", "serve my fine-tuned model", or any request that involves making a model available for inference on AWS. Use this even when the user is vague (e.g. "I just want to get this running on AWS, you figure it out"). Works for text-generation LLMs, embedding models, rerankers, classifiers, text-to-image / diffusion models — picks the right serving stack and chooses between real-time and async inference. This is the entry-point skill for SageMaker deployment work — it asks clarifying questions, picks a deployment pathway, and coordinates the other deployment skills.

Pick the right serving container for a SageMaker model deployment and find its current image URI. Use this skill whenever about to deploy a model to a SageMaker endpoint and an image URI needs to be chosen — including when the user says "deploy this LLM", "host this HuggingFace model", "serve this fine-tuned model", "deploy this embedding model", "host a reranker", "serve a sentence-transformers model", or when about to hardcode any container URI in deployment code. HuggingFace-curated Deep Learning Containers are ALWAYS preferred: HuggingFace vLLM (LLMs and generative rerankers), HuggingFace vLLM-Omni (multimodal), TEI (embeddings/cross-encoder rerankers), HF Inference Toolkit (other transformers). Generic images (AWS vLLM, DJL-LMI, SGLang) are used only when no HuggingFace image is compatible — never merely because they carry a newer version. Never hardcode a container URI from memory and never default to TGI. Prevents stale-image failures and wrong-region URIs.

Ensure a usable SageMaker execution role exists before deploying or training. Use this skill whenever about to create a SageMaker endpoint, model, training job, or any resource that requires an execution role. Use it especially when the user has not provided a role ARN explicitly, when scripts are about to call `iam:CreateRole`, or when an AccessDenied error mentions an IAM action. Never blindly call `iam:CreateRole` — always check for existing roles first. This skill prevents the most common SageMaker deployment failure: trying to create IAM resources from an SSO principal that has no IAM write permissions.

Discover the user's local AWS context (active profile, region, account ID, caller identity) at the start of any AWS task. Use this skill before any other AWS work — deploying to SageMaker, creating resources, calling AWS APIs, or anything that touches an AWS account. Use it especially when the user has not specified a region or profile explicitly, when they say things like "use my AWS account", "deploy to AWS", "use my profile", or when about to make any AWS CLI or SDK call. Never guess the region or account ID — always use this skill to read it from the local configuration first.

Build, deploy, and maintain applications on Hugging Face Spaces — Gradio / Docker / Static SDKs, ZeroGPU and dedicated hardware, model loading, debugging, buckets, inference providers, community grants. Use whenever the user asks to create or host an app on Hugging Face, port code onto ZeroGPU, fix a Space that won't build or run, or otherwise work with `hf spaces …`, `@spaces.GPU`, Space README frontmatter, or the `spaces` Python package.