Feedback on your aws-solution-architect skill

[RichardHightower]

Checked out your AWS solution architect skill and noticed it's tackling the infrastructure-as-code problem space—curious how you're thinking about the balance between prescriptive guidance and letting users make their own architectural tradeoffs, especially at a 58/100 where some of those fundamentals might still be settling.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 58/100, which lands this in F territory. Based on Anthropic's skill evaluation standards, your strongest area is Spec Compliance (11/15), but you're really struggling with Progressive Disclosure Architecture (13/30) and Writing Style (5/10). The gap between what you're covering and how you're organizing it is the main lever here.

What's Working Well

• Solid frontmatter - Your YAML metadata is clean and spec-compliant with all required fields present
• Consistent AWS terminology - You're using service names accurately throughout (Lambda, DynamoDB, EventBridge, etc.), which makes the content technically sound
• Multiple architecture patterns - You've included serverless, containerized, and hybrid approaches, giving users actual options to consider

The Big One: All 345 Lines in One File

This is crushing your PDA score. Right now, everything—architecture patterns, best practices, service guides, security considerations—is crammed into a single monolithic SKILL.md. That violates progressive disclosure entirely.

The fix: Split into a reference-based structure:

• Keep SKILL.md under 100 lines with just the overview and capabilities
• Create references/architecture-patterns.md for serverless/container/hybrid patterns
• Create references/service-selection-guide.md for choosing DynamoDB vs. RDS, etc.
• Create references/security-best-practices.md for IAM, encryption, compliance
• Create references/cost-optimization.md for reserved instances, spot pricing, etc.

This alone gets you +7 points and fixes your navigation (no TOC needed for a 50-line main file).

Other Things Worth Fixing

1. Add explicit trigger phrases to your description - Right now it says "Expert AWS solution architecture for startups..." but doesn't tell Claude when to activate. Add: Triggers: "design AWS architecture", "serverless stack", "CloudFormation template", "AWS cost optimization". Easy +2 points.

2. Ditch the marketing language - "Expert," "comprehensive," "modern cloud-native patterns," "proven pattern" all read like a sales pitch. Replace with technical specifics: "Provides architecture design for Lambda, DynamoDB, EventBridge, and infrastructure-as-code templates." Nets you +3 points on writing style.

3. Add actionable workflows - Your patterns describe what (API Gateway + Lambda) but not how. Add numbered steps: 1. Define API schema 2. Create Lambda function 3. Configure API Gateway triggers 4. Deploy with CDK 5. Test endpoints 6. Monitor CloudWatch logs. Gets you +3 points on utility.

4. Include validation steps - No feedback loops anywhere. After each architecture recommendation, add: "To validate: run aws cloudformation describe-stacks, check stack status, test API endpoints with curl, review CloudWatch logs." This adds the run→check→fix pattern your skill is missing (+3 points).

Quick Wins

Most impactful fixes (in order):

1. Refactor into reference-based structure (+7 points) ← Do this first
2. Add trigger phrases to description (+2 points)
3. Strip marketing language, use technical specifics (+3 points)
4. Add numbered implementation workflows (+3 points)
5. Include validation/verification steps (+3 points)

These five changes alone get you from 58 → 79, moving you into C territory. The reference refactor is the heaviest lift but has the biggest payoff.

---

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.