bash-defensive-patterns

I took a look at your bash-defensive-patterns skill and wanted to share some thoughts.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 66/100, which puts you in D territory. This is evaluated against Anthropic's best practices for agentic skills. Your strongest area is Spec Compliance (12/15) — the metadata and naming are solid. But Progressive Disclosure Architecture (14/30) is where you're losing the most points. The skill's a flat 534-line file that loads entirely every time, which isn't efficient for how agentic systems work.

What's Working Well

• Solid metadata: Your triggers ('production-grade scripts', 'CI/CD pipelines', 'fault tolerance') are specific and searchable. The skill name follows conventions perfectly.
• Comprehensive coverage: You actually cover the critical patterns — strict mode, error trapping, signal handling. This addresses a real gap in defensive Bash practices.
• Clear organization: Numbered patterns and section headers make manual navigation reasonable, even if it could be better.
• Practical scope: Good balance between providing complete patterns and letting users adapt them to their needs.

The Big One: Progressive Disclosure Architecture

This is your biggest opportunity. Right now, all 10 patterns plus advanced techniques live in one massive file. Users get everything whether they need it or not.

Here's the fix: Break it into a layered structure:

• Create a references/ directory with pattern-specific files: strict-mode.md, error-trapping.md, argument-parsing.md, cleanup-patterns.md, etc.
• Reduce SKILL.md to an 80-line overview that summarizes each pattern and links to the references
• This way, the main file loads fast, but users can dive into specific patterns on demand

Why this matters: Agentic systems load skill content for every activation. A 534-line file means you're burning tokens every single time, even when someone just needs one pattern. Splitting it up gets you +10 points and makes your skill way more efficient.

Other Things Worth Fixing

1. Add a table of contents — Files over 100 lines should have one. Just add a quick ## Contents with links to your main sections after the frontmatter. (+1 point, takes 30 seconds)

2. Cut the boilerplate in examples — Patterns 1-10 each show full 20-40 line scripts with repeated set -Eeuo pipefail and trap setup. Reference a shared defensive template instead and show just the diff. This cuts verbosity in half and scores better on writing conciseness. (+2 points)

3. Add validation workflow — No feedback loop currently. Add a "Verify Your Script" section with actual commands: shellcheck -x script.sh and bash -n script.sh. Users need to know how to test what they built. (+2 points)

4. Include a quick-start template — At the end, provide one minimal script combining core patterns (strict mode + logging + arg parsing + cleanup). Users shouldn't have to mentally assemble pieces. (+1 point)

Quick Wins

1. Split into references — Biggest impact, biggest effort. Gets you 10 points toward closing that PDA gap.
2. Add TOC + cut example boilerplate — Easy wins, 3 points combined, takes maybe 20 minutes.
3. Validation section + starter template — Another 3 points, rounds out the skill nicely.

You've got solid foundational content here. The main issue is packaging — make it lean, make it layered, and make it easy to verify. Fix these and you're looking at 80-85 range territory.

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.