stelow-workflow-codebase-critique

Codebase Critique

Focus: Critically analyze source code structure — architecture, data flow, API contracts, performance, theming, responsive patterns, and AI slop in code. Input: Source code directory. Output: Classified gap report (🚨/🤔/🔎) with recommendations.

Tools: See ../stelow-workflow-orchestrator/references/cli-tools/subagents.md for subagent patterns.

Standalone awareness: inside stelow, quality gates critique depth (experimental → light, production → full). Standalone defaults to production quality (full single-reviewer). Works with any source directory — no stelow dependency for the audit logic.

Overview

Structural codebase critique — evaluates architecture, data flow, API contracts, performance, theming, responsive patterns, and AI slop in code.

This skill performs a source code audit using checklists focused on technical and structural aspects (non-visual):

Dimension	What it evaluates
🏗️ Architecture	Module structure, dependency flow, coupling
🔄 Data Flow	Call chains, state management, event propagation
📡 System Contracts	API definitions, error handling, logging
⚡ Performance	Bundle size, re-renders, lazy loading
🎨 Theming	Design tokens, dark mode support
📱 Responsive	CSS breakpoints, media queries
🤖 AI Slop in Code	Over-generated patterns, redundant code, boilerplate

Important: This skill is for analyzing source code of components and logic. If you need visual UI auditing (accessibility, design, UX), use stelow-workflow-ux-critique em Codebase mode.

Quality Gate (auto-skip for small scopes)

Before running codebase critique, check if quality warrants it. Codebase critique is for structural analysis — if the scope is 1 file, the value is minimal.

# Read quality from stelow context or env var; default production (via canonical helper).
WF_DIR="$(ls -td .stelow/*/*/ 2>/dev/null | head -1)"
# shellcheck disable=SC1091
source "$(dirname "${BASH_SOURCE[0]:-$0}")/../../stelow-workflow-orchestrator/references/cli-tools/read-config.sh"
QUALITY="${QUALITY:-$(stelow_read_quality)}"
DIFF_FILES=$(git diff --name-only HEAD~1 2>/dev/null | wc -l | tr -d ' ')
# Entity-level gate when sem is available (files undercount refactors that touch few files broadly)
DIFF_ENTITIES=$(sem diff --format json HEAD~1 2>/dev/null | python3 -c "import sys,json; print(len(json.load(sys.stdin).get('entities', [])))" 2>/dev/null || echo "$DIFF_FILES")