reasoningbank-intelligence

I took a look at your reasoningbank-intelligence skill and wanted to share some thoughts.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 63/100, which puts you in D territory. The good news? You've got solid utility (14/20) - the core idea of adaptive pattern recognition for AI agents addresses a real gap. The rough spots are in Progressive Disclosure Architecture (15/30) and Spec Compliance (10/15). This evaluation is based on Anthropic's best practices for agent skills.

What's Working Well

• Practical utility - The ReasoningBank concept solves a real problem; agents learning from their own experiences is genuinely useful
• Good trigger phrases - Your metadata picks up on "self-learning," "meta-cognitive," and "adaptive" which helps discoverability
• Configuration flexibility - Exposing learningRate, threshold, and updateFrequency gives users real control
• Solid examples - You've got concrete TypeScript code showing how to use this, not just abstract descriptions

The Big One: You're Putting Too Much in One File

Right now, all 202 lines are crammed into SKILL.md. That's verbose and burns tokens unnecessarily. Here's what's holding you back: no references/ directory structure at all.

The fix: Split this into layers:

• Keep SKILL.md to ~50 lines (just intro + minimal Quick Start)
• Create references/api-reference.md for all those TypeScript code blocks
• Create references/patterns.md for the best practices section
• Create references/troubleshooting.md for edge cases

This alone gets you +8 points. Your file structure should look like:

reasoningbank-intelligence/
├── SKILL.md (just overview + quick start)
├── references/
│   ├── api-reference.md
│   ├── patterns.md
│   └── troubleshooting.md

Other Things Worth Fixing

1. Name convention mismatch - Your frontmatter says name: ReasoningBank Intelligence but it should be lowercase: name: reasoningbank-intelligence. Quick fix, +2 points.

2. Missing table of contents - At 202 lines, add a TOC right after the description. Takes 30 seconds, helps users navigate. +2 points.

3. No workflow checklist - Your "Best Practices" section is just bullets. Add a validation checklist like:

   - [ ] ReasoningBank initialized with persist: true
   - [ ] First experience recorded successfully
   - [ ] Metrics show experiences > 0
   

   This makes it actually usable. +2 points.

4. Step-by-step workflow is missing - Your Quick Start just dumps code. Break it into numbered steps: "Step 1: Install," "Step 2: Initialize," "Step 3: Record First Experience." +3 points.

Quick Wins (by impact)

• Restructure into references/ subdirectories (+8)
• Fix name convention to lowercase (+2)
• Add TOC and numbered workflow steps (+5)
• Add validation checklist (+2)

That's +17 points if you nail these, bumping you to 80/100 (solid B 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.