Skip to content
Get packs

SageMaker Production Endpoint Defaults

Deploys AWS SageMaker endpoints with autoscaling, CloudWatch alarms and tagging on by default

Source Download skill
Repository
huggingface/skills
Stars
11,253
Licence
Apache-2.0
Files
8 files

What it does

When an agent is about to create an AWS SageMaker endpoint (a hosted server that runs an AI model), this skill makes it use a set of ready-made Python scripts instead of a bare create call. The scripts build the model, endpoint config, scaling rules, CloudWatch alarms and resource tags together, so the endpoint can scale with traffic, report failures, and be found and cleaned up later. The skill also requires a smoke test (a real test request plus a scan of logs for crash messages) before the agent reports success.

Great for

  • Engineers deploying a vLLM or embedding model to SageMaker who want a repeatable, production-style setup
  • Teams that need endpoints that can scale to zero between batches or idle periods
  • People who already have an image URI, IAM role and AWS region chosen by other skills in the same workflow

Skip it if

  • Users without an AWS account or without permission to create SageMaker, IAM and CloudWatch resources, since the scripts create billable cloud resources
  • Anyone expecting a one-click cost-free test: endpoints, alarms and optional data capture to S3 incur ongoing charges
  • Non-SageMaker deployments, since the skill is specific to AWS SageMaker APIs

Try asking

  • “Deploy a Qwen model on an ml.g5.xlarge SageMaker endpoint with autoscaling and alarms”
  • “Create a scale-to-zero real-time endpoint for a small model using inference components”
  • “Set up an async SageMaker endpoint that scales to zero between batch jobs”
  • “Tear down a SageMaker endpoint created by these scripts and check the resources are gone”
Step-by-step guide: how to use SageMaker Production Endpoint DefaultsInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 8 original files, the Apache-2.0 licence and an attribution note.

    hf-cloud-sagemaker-production-defaults.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip hf-cloud-sagemaker-production-defaults.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip hf-cloud-sagemaker-production-defaults.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Package model weights in your own S3 bucket and pass --model-s3-uri so the container does not download from the Hub at runtime; review the scripts before running them with real credentials
  • Start with a small instance type and a short test, then tear down with the printed teardown command to avoid ongoing charges
  • If a diagnostic call is denied by your IAM role, the skill is designed to note that and continue with the checks that work rather than assume all is well

What's inside

  • references/deployment-template.md
  • scripts/_common.py
  • scripts/deploy_async.py
  • scripts/deploy_ic.py
  • scripts/deploy.py
  • scripts/invoke_endpoint.py
  • scripts/teardown.py
  • SKILL.md

Redistributed unmodified from huggingface/skills under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.