I checked out your AWS Solution Architect skill and noticed the grading came in at 66/100 — there's a solid foundation here, but I'm curious what trade-offs you made between comprehensive coverage and keeping the documentation maintainable at this scope.
Links:
The TL;DR
You're at 66/100, D territory — based on Anthropic's best practices for agentic skills. Your strongest area is Spec Compliance (11/15), but Writing Style is dragging you down hard at just 4/10. The core issue isn't that your content is bad — it's that you're treating this like a reference manual when it should be a focused, actionable guide.
What's Working Well
- Solid spec foundation — Your frontmatter is valid YAML with all required fields, and you're using proper naming conventions (hyphen-case). That's the baseline done right.
- Real AWS expertise — The architecture patterns and service selection show you actually know your stuff. Cost breakdowns and trade-offs feel grounded in real experience.
- Helpful layering attempt — Your HOW_TO_USE.md file shows you understand that progressive disclosure matters; you just need to push that idea further into the main skill.
The Big One: SKILL.md is 345 Lines of Bloat
This is your main bottleneck. You've got 6 detailed architecture patterns, extensive best practices, service catalogs, and cost estimates all crammed into the primary file. That's not a skill guide — that's an AWS reference manual.
Why it matters: Every line dilutes the signal. When someone asks Claude to use your skill, they get 345 lines to parse when they probably need 10-15 actionable steps. You're burying the workflow in encyclopedia content.
The fix: Move those 6 architecture patterns to references/architecture_patterns.md. Keep only a 2-3 sentence teaser in SKILL.md pointing to it. Do the same with service comparisons and best practices — they belong in references/, not the main file. Your actual workflow (gather requirements → design → generate IaC → validate → deploy) should be the skeleton of SKILL.md, with references hanging off it.
Impact: +8 points just from better token economy and cleaner navigation.
Other Things Worth Fixing
-
No trigger phrases in the description — Your frontmatter description mentions 'serverless, scalable' but doesn't include the actual phrases that should activate this skill. Add: "Use when asked to 'design serverless architecture', 'create CloudFormation template', 'optimize AWS costs'". +2 points.
-
Missing workflow steps — Replace the encyclopedic "Capabilities" section with a numbered workflow: (1) Gather requirements, (2) Run architecture_designer.py, (3) Review and cost-check, (4) Generate IaC, (5) Validate, (6) Deploy. Right now it feels like a reference, not a process. +5 points.
-
Cut the marketing language — Phrases like "Expert" and "comprehensive" read like a job posting, not technical guidance. Be objective: "Design AWS architectures for startups using serverless patterns and IaC templates." +3 points.
-
Second-person voice inconsistency — You slip between imperative ('use Lambda') and descriptive ('you'll receive diagrams'). Pick one and stick with it throughout both files.
Quick Wins
- Move 6 patterns to references/ folder — biggest token win, +8 points
- Add trigger phrases to description — clear activation signal, +2 points
- Structure SKILL.md around workflow steps — makes it actually usable, +5 points
- Strip marketing language — fixes writing style, +3 points
That's +18 points possible if you hit all four. You'd jump from 66 to 84 just by refocusing the signal.
Checkout your skill here: SkillzWave.ai | SpillWave 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.
I checked out your AWS Solution Architect skill and noticed the grading came in at 66/100 — there's a solid foundation here, but I'm curious what trade-offs you made between comprehensive coverage and keeping the documentation maintainable at this scope.
Links:
The TL;DR
You're at 66/100, D territory — based on Anthropic's best practices for agentic skills. Your strongest area is Spec Compliance (11/15), but Writing Style is dragging you down hard at just 4/10. The core issue isn't that your content is bad — it's that you're treating this like a reference manual when it should be a focused, actionable guide.
What's Working Well
The Big One: SKILL.md is 345 Lines of Bloat
This is your main bottleneck. You've got 6 detailed architecture patterns, extensive best practices, service catalogs, and cost estimates all crammed into the primary file. That's not a skill guide — that's an AWS reference manual.
Why it matters: Every line dilutes the signal. When someone asks Claude to use your skill, they get 345 lines to parse when they probably need 10-15 actionable steps. You're burying the workflow in encyclopedia content.
The fix: Move those 6 architecture patterns to
references/architecture_patterns.md. Keep only a 2-3 sentence teaser in SKILL.md pointing to it. Do the same with service comparisons and best practices — they belong inreferences/, not the main file. Your actual workflow (gather requirements → design → generate IaC → validate → deploy) should be the skeleton of SKILL.md, with references hanging off it.Impact: +8 points just from better token economy and cleaner navigation.
Other Things Worth Fixing
No trigger phrases in the description — Your frontmatter description mentions 'serverless, scalable' but doesn't include the actual phrases that should activate this skill. Add: "Use when asked to 'design serverless architecture', 'create CloudFormation template', 'optimize AWS costs'". +2 points.
Missing workflow steps — Replace the encyclopedic "Capabilities" section with a numbered workflow: (1) Gather requirements, (2) Run architecture_designer.py, (3) Review and cost-check, (4) Generate IaC, (5) Validate, (6) Deploy. Right now it feels like a reference, not a process. +5 points.
Cut the marketing language — Phrases like "Expert" and "comprehensive" read like a job posting, not technical guidance. Be objective: "Design AWS architectures for startups using serverless patterns and IaC templates." +3 points.
Second-person voice inconsistency — You slip between imperative ('use Lambda') and descriptive ('you'll receive diagrams'). Pick one and stick with it throughout both files.
Quick Wins
That's +18 points possible if you hit all four. You'd jump from 66 to 84 just by refocusing the signal.
Checkout your skill here: SkillzWave.ai | SpillWave 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.