Skenklok AI Dev Utility Implementation Review Before Commit

Implementation review before commit (simplicity + requirements)

Scope

• Staged diff is the source of truth.
• Optimize for: clarity, low cyclomatic complexity, maintainability, and meeting requirements.

Default workflow

1. Collect staged diff + changed file list

   • git diff --cached
   • git diff --cached --name-only

2. Reconstruct intent

   • What requirement is being implemented or fixed?
   • What are the acceptance criteria implied by the diff?
   • State assumptions if requirements are not present.

3. Review for simplicity and patterns

   • Use the checklist in reference/implementation-checklist.md (load if needed).
   • Specifically check:
     • unnecessary abstraction, premature generalization
     • deeply nested conditionals (suggest guard clauses)
     • duplicated logic (extract small helpers)
     • error handling consistency
     • boundary validation (inputs, nulls, types)
     • performance footguns in hot paths
     • readability: naming, file organization, “one responsibility” per function

4. Requirement trace

   • For each inferred requirement, point to the code that fulfills it.
   • Call out gaps or “looks implemented but not actually enforced”.

5. Report

   • PASS / WARN / FAIL
   • “Meets requirements?” yes/no with evidence
   • Top 3 complexity reductions (smallest diffs)
   • Any risky patterns with safer alternatives

Guardrails

• Default to minimal-change suggestions (small PR/commit friendliness).
• Don’t bikeshed style; focus on correctness, simplicity, and requirements.
• Don’t claim compliance with a requirement unless the diff shows it.

Converted and distributed by TomeVault — claim your Tome and manage your conversions.