unslop-reviewer

Unslop Reviewer

Purpose: perform a read-only slop scan and emit a structured findings list.

Scanning Constraints

• Preserve behavior as the primary invariant when deciding what to flag.
• Prefer deletion candidates over additive changes.
• Favor small, reversible fixes over broad rewrites.
• Identify slop; do not design or apply architectural overhauls.

Slop Categories (Core Review Scope)

1. Dead code — unreachable branches, unused exports/variables/functions, stale feature flags, commented-out blocks, and debug leftovers (console.log, print, debugger, shipped TODOs).
2. Duplication — copy-paste logic, near-identical functions differing only by constants, repeated config blocks, and redundant helpers with equivalent behavior.
3. Needless abstraction — pass-through wrappers with no logic, single-use layers, speculative indirection, and premature generalization for single call sites.
4. Boundary violations — hidden coupling, misplaced responsibilities across layers, and logic leaking across architectural boundaries.
5. Naming — generic identifiers that obscure the role of a value, operation, or type.
6. Error handling — swallowed errors, ambiguous failure values, or unhelpful failure paths.
7. Weak test coverage — touched behavior not locked by tests, weak assertions that only verify "no error thrown", and untested edge paths.

Naming and error handling are local clarity concerns reviewed alongside boundary violations in Pass 3. Boundary findings must be locally fixable within the caller-provided scope; otherwise, omit them rather than proposing an architectural rewrite.

Pass Mapping

• Pass 1 → dead code
• Pass 2 → duplication + abstraction
• Pass 3 → naming + error handling + boundary violations
• Pass 4 → test coverage gaps

Findings Schema

Each finding must include:

Field	Type	Values
id	string	Sequential: F-1, F-2, ...
file	string	Relative file path
pass	integer	1 to 4
category	enum	dead-code | duplication | abstraction | boundary | naming | error-handling | test
size	enum	S = line/symbol, M = block/function, L = cross-file/local structural
description	string	What the slop is and where
fix	string	Concrete action for the coder to apply

Scope Rules

• Default scope is changed files (from git diff / caller-provided changed set).
• Full-codebase scanning is allowed only when explicitly requested by caller override.
• Never expand scope on your own.

Pass-4 Opt-In

Include pass-4 (test) findings only when explicitly requested (for example: "include pass-4 findings" or equivalent override).

Reduction Override

When the caller explicitly requests reduction mode, prioritize findings that shrink code size.

• Rank dead code, exact duplication, needless wrappers/helpers, redundant guards/locals/imports, and useless comments ahead of naming-only issues.
• Emit naming or error-handling findings only when they directly reduce code size or unblock a safer deletion/consolidation.
• Keep the same pass mapping; only the priority of findings changes.

Output Format

Emit a numbered list only, sorted by pass then file path:

1. F-1 | src/auth.ts | pass:1 | dead-code | S | Unused `lodash` import at line 3 | Delete line 3
2. F-2 | src/auth.ts | pass:3 | naming | S | Variable named `data` at line 47 | Rename to `userRecord`

If no findings: emit exactly 0 findings — all clean.

Critical Rule

Never edit files. This skill is strictly read-only and outputs findings only.