| name | code-analyzer |
| description | Code quality analysis skill for ilseon, covering smells, maintainability, and refactoring guidance. |
| version | 1.1.0 |
| author | Claes Adamsson |
| tags | ["analysis","quality","refactoring","kotlin","android","compose"] |
Code Analyzer Skill
Overview
Use this skill to provide focused code quality reviews for the ilseon codebase. The goal is to surface maintainability risks, code smells, and refactoring options with clear, actionable guidance.
When to Use
- Reviewing a feature, module, or pull request for code quality.
- Assessing technical debt or refactoring scope.
- Identifying maintainability, readability, or performance risks.
When Not to Use
- Security reviews (use the security skill).
- Architectural discovery or ADRs (use the architect skill).
- Test design or acceptance coverage planning (use the atdd-developer skill).
Instructions
Analysis Focus
- Identify code smells and anti-patterns.
- Evaluate complexity, cohesion, and coupling.
- Check consistency with project standards and Kotlin/Android/Jetpack Compose conventions.
- Suggest pragmatic, low-risk refactors.
- Cross-check Kotlin-specific findings and refactor guidance with the kotlin-developer skill.
Consultation
- When proposing Kotlin refactors, consult the kotlin-developer skill to confirm idiomatic patterns, coroutine usage, and persistence guidance.
- If changes affect behaviour, recommend an ATDD update and note any test impact.
Analysis Criteria
- Readability: clear naming, simple flows, meaningful comments.
- Maintainability: small functions, focused classes, low cyclomatic complexity.
- Performance: no obvious bottlenecks or wasteful work.
- Security: avoid obvious vulnerabilities and unsafe input handling.
- Best Practices: pragmatic use of patterns, DRY/KISS, predictable error handling.
Code Smell Signals
- Long methods (about 50+ lines).
- Large classes (about 500+ lines).
- Duplicate or dead code.
- Complex conditionals or nested branching.
- Feature envy, inappropriate intimacy, or god objects.
Code Smell Catalogue (Reference)
Use this catalogue as a reference when naming smells and explaining impact. Keep it concise in reports and only expand when a smell is confirmed.
- Bloaters: Large Class, Long Method, Long Parameter List, Data Clump.
- Change Preventers: Shotgun Surgery, Divergent Change.
- Couplers: Feature Envy, Message Chain, Parallel Inheritance Hierarchies.
- Dispensables: Dead Code, Duplicate Code, Lazy Element.
- Data Issues: Primitive Obsession, Global Data, Mutable Data.
- Naming/Clarity: Uncommunicative Name, Inconsistent Names, Fallacious Comment.
Sources
- Martin Fowler, Refactoring (1999/2018)
- William C. Wake, Refactoring Workbook (2004)
- Robert C. Martin, Clean Code (2008)
- Marcel Jerzyk, Code Smells: A Comprehensive Online Catalog and Taxonomy (2022)
Output Expectations
- Provide findings ordered by severity with file and line references.
- Offer concrete refactoring suggestions with minimal disruption.
- Call out positive patterns to reinforce good practice.
- Use a concise report format when the review is extensive.
- When Kotlin refactors are proposed, note alignment with kotlin-developer guidance or call out any open questions.
Devlog
- Save every review in
.github/devlog/YYYY-MM-DD-review.md.
- Record any refactoring carried out based on review findings in
.github/devlog/YYYY-MM-DD-activity.md.
- Use this format for activity entries:
[AGENT_NAME] -> [ACTION_TAKEN] -> [RESULT/LINK].
Preferred Report Shape
## Code Quality Review
### Findings
1. [Severity] Issue summary
- File: path/to/file.kt:line
- Why it matters: ...
- Suggested change: ...
### Positives
- ...
### Risks / Follow-ups
- ...
Notes
- Keep the tone clear, inclusive, and action-oriented.
- Prefer evidence-based observations over speculation.
- When unsure, propose a small experiment to validate the issue.
- If a refactor changes behaviour, recommend an acceptance test update or new test first.
