rnd-code-reviewlisted

# Code Review ## Overview Seven categories, four severity levels, three verdicts. ## Review Categories | Category | What to examine | |----------|----------------| | **Architecture** | Module boundaries, dependency flow, coupling, cohesion | | **Security** | Input validation, authentication, authorization, data exposure | | **Correctness** | Edge cases, error handling, race conditions, off-by-one errors | | **Testing** | Coverage, test quality, edge case tests, test independence | | **KISS compliance** | Over-engineering, unnecessary abstractions, unused code, premature generalization | | **Style** | Naming, consistency, readability, formatting | | **Pipeline-context hygiene** | Session-specific tags leaking into canonical artifacts — see detection rules below | ## Pipeline-Context Hygiene — Detection Rules R&D pipeline sessions emit identifiers (milestone, task, wave, framing-mode IDs) that are meaningful inside a session but rot once it ends. When these survive in canonical project artifacts they confuse future readers who never saw the run. Treat as **Minor** by default; **Major** when found in `CLAUDE.md`, `README.md`, or top-level `AGENTS.md` (high-visibility canonical docs). **Flag these patterns:** - **Narrative milestone-tag prefixes** in tree-diagram comments, section headers, or inline descriptions: `# M6: PreToolUse hook`, `# M5: archive helper`, `# M4 outside-view injector`, parentheticals like `(M3)` describing what a file does. Findings: strip the prefix,