code-reviewer

I was digging through the grading results and noticed your code-reviewer skill came back at 40/100—there's clearly solid intent here, but the execution is falling short of what this tool could be doing. What I'm curious about is whether the issue is in how the reviews are structured or something deeper in the analysis logic?

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 40/100, which lands you in F territory. This is based on Anthropic's Progressive Disclosure Architecture standards for agentic skills. Your strongest area is Spec Compliance (12/15)—the metadata and frontmatter are clean. But Utility (4/20) and Writing Style (3/10) are really dragging things down. The skill promises a "complete toolkit" but the actual implementation is mostly empty templates.

What's Working Well

• Valid YAML frontmatter with all required fields present and correct naming conventions
• Good trigger coverage in the metadata—you're targeting reviewers who want to check PRs, get code feedback, and scan for quality issues
• Proper reference file structure—you've got the framework set up for common_antipatterns.md, coding_standards.md, and code_review_checklist.md, which shows you thought about layering

The Big One: Empty Scripts Kill Utility

Your three scripts (pr_analyzer.py, code_quality_scanner.py, review_formatter.py) are just boilerplate with placeholder comments like // Implementation details and self.results['status'] = 'success'. This is the anchor around your neck.

Why it matters: Reviewers need actual logic—linting checks, complexity analysis, security scanning patterns. Right now, you're promising a toolkit but delivering a shell.

The fix: Either implement real review logic (check for SOLID violations, hardcoded credentials, error handling patterns) or remove the scripts entirely and focus on reference materials. If you go the reference route, turn your three checklist files into actual checklists with checkboxes and specific criteria (e.g., [ ] Functions have single responsibility, [ ] No console.logs in production code). This alone could bump you +5-7 points.

Other Things Worth Fixing

1. Generic reference content — Your three reference files are identical templates with no code-review-specific patterns. Replace with real antipatterns (e.g., nested ternaries, missing null checks, performance red flags). +7 points potential

2. Marketing fluff — Lines like "Complete toolkit" and "Comprehensive guide" don't describe what the skill does. Change to concrete actions: "Analyzes PR diffs for changed files, checks commit message format, identifies review complexity." +5 points potential

3. Terminology inconsistency — You're using both "Pr Analyzer" and "PR analyzer" throughout. Standardize on "PR Analyzer" for tool names. +2 points potential

4. Missing navigation — Your SKILL.md is 210 lines without a table of contents. Add a quick TOC after the intro so users can jump to sections. +2 points

Quick Wins

• Replace empty scripts with either real code or remove them
• Convert reference files from templates to actual code review patterns and checklists
• Strip marketing language and replace with specific, actionable descriptions
• Add table of contents to SKILL.md
• Standardize capitalization across all tool names

These fixes could realistically get you to 65-75/100 with relatively straightforward work.

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.