Skip to content

Feedback on your using-localstack skill #1

@RichardHightower

Description

@RichardHightower

I took a look at your using-localstack skill and wanted to share some thoughts.

Links:

The TL;DR

You're at 88/100, solid B territory. This is based on Anthropic's Claude Skills best practices. Your strongest area is Utility (19/20)—the skill genuinely solves real problems for local AWS development. The weakest is Spec Compliance (11/15), which is fixable low-hanging fruit.

What's Working Well

  • Progressive Disclosure Architecture (28/30) is nearly perfect. Your SKILL.md is only 94 lines but packed with signal—no fluff. The layered structure with 10 reference files, examples directory, and specific troubleshooting guides is exactly what we want to see.

  • Utility (19/20) is your standout. You're addressing real gaps: local AWS development without paying for cloud cycles, deterministic CI/CD testing, and IAM debugging. The skill knows its constraints too (when NOT to use LocalStack).

  • Trigger terms are excellent: awslocal, LocalStack, Lambda, S3, DynamoDB, Terraform, CDK, CI/CD—developers will find this naturally when they search.

The Big One: Add Trigger Phrases to Description

Your description is accurate but missing trigger phrases, and that's costing you 2 points in Spec Compliance. Right now it reads:

description: Build, test, and debug AWS-native systems locally with LocalStack (Community/Pro)...

What you need:

description: Use this skill when asked to "set up LocalStack", "run LocalStack locally", "LocalStack debugging", or "AWS local testing".

Why it matters: Trigger phrases help Claude understand when to invoke your skill. Without them, developers have to stumble into it by accident. This is a 2-point swing and takes 30 seconds to fix.

Other Things Worth Fixing

  1. Voice consistency in troubleshooting — A few spots have second-person "you" (like "If you encounter bugs..."). Rewrite as "When encountering bugs..." to stay imperative. Minor, but it's worth 1 point.

  2. Add TOCs to longer reference filesdocker-desktop-troubleshooting.md and rancher-desktop-troubleshooting.md are solid but would benefit from a table of contents at the top, especially if they hit 100+ lines. Makes navigation faster.

  3. Metadata consistency — You're using gerund-style naming, which is great. Just make sure all optional fields in your frontmatter are used where they add value (like allowed-tools if you're calling specific commands).

Quick Wins

  • Add trigger phrases to description (+2 points) — Biggest bang for your buck
  • Clean up voice in troubleshooting (+1 point) — Quick pass through references
  • Add TOCs to files over 80 lines — Better UX, minimal effort
  • The architecture and examples are already solid—you're not building from scratch

Checkout your skill here: [SkillzWave.ai](https://skillzwave.ai) | [SpillWave](https://spillwave.com) We have an agentic skill installer that install skills in 14+ coding agent platforms. Check out this guide on how to improve your agentic skills.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions