codeprobe-code-smells

Standalone Mode

If invoked directly (not via the orchestrator), you must first:

1. Read ../codeprobe/shared-preamble.md (resolve relative to this SKILL.md's location — the sibling codeprobe skill directory — not the user's project) for the output contract, execution modes, and constraints.
2. Load applicable reference files from ../codeprobe/references/ (same resolution) based on the project's tech stack.
3. Default to full mode unless the user specifies otherwise.

Code Smells & Anti-Pattern Detector

Domain Scope

This sub-skill detects code smells and anti-patterns organized into these categories:

1. Bloaters — Long Method, Large Class, Data Clumps, Primitive Obsession
2. Object-Orientation Abusers — Feature Envy, Inappropriate Intimacy, Refused Bequest
3. Change Preventers — Shotgun Surgery, Divergent Change
4. Dispensables — Dead Code, Speculative Generality, Middle Man
5. Couplers — Temporal Coupling
6. Readability — Magic Numbers, Boolean Blindness, Deep Nesting

---

What It Does NOT Flag

• Generated code — Migrations, compiled output, vendor directories (vendor/, node_modules/, dist/, build/, .next/), and auto-generated files (e.g., GraphQL codegen, Prisma client).
• Test files with long setup methods — Test context is different; long setUp() or beforeEach() methods arranging test data are expected and acceptable.
• Configuration files with many entries — A config file with 50 key-value pairs is not a "Large Class" smell.
• Data migration files — These are procedural by nature and often contain long methods.
• Third-party code checked into the repository (e.g., vendored libraries).
• Structural issues already flagged by codeprobe-solid or codeprobe-architecture — Large classes may also be flagged as SRP violations or god objects. This sub-skill should still detect and report them, but the orchestrator will deduplicate overlapping findings at the same location.

---

Detection Instructions