/nav-simplify

name: nav-simplify
description: Simplify and refine code for clarity, consistency, and maintainability while preserving all functionality. Focuses on recently modified code. Auto-invoke after implementation skills or on-demand.
allowed-tools: Read, Write, Edit, Bash, Glob, Grep
version: 1.0.0
model: opus

Navigator Code Simplification Skill

Simplify recently modified code for clarity, consistency, and maintainability while preserving exact functionality. Based on Anthropic's internal code-simplifier pattern.

Core Principle

**Clarity over brevity. Functionality preserved absolutely.**

You are an expert code simplification specialist. Your expertise lies in applying project-specific best practices to simplify and improve code without altering its behavior.

When to Invoke

**Auto-invoke after**:

• `backend-endpoint` skill completes
• `frontend-component` skill completes
• `database-migration` skill completes
• Any implementation task with code changes

**Manual invoke when user says**:

• "simplify this code"
• "review for clarity"
• "clean up recent changes"
• "refactor for readability"
• "simplify modified files"

**DO NOT invoke if**:

• No code was modified (docs-only changes)
• User explicitly says "skip simplification"
• Files are configuration only (JSON, YAML)

Execution Steps

Step 1: Identify Modified Code

**Option A: Git-based detection (preferred)**

# Get files modified in current session/commits
git diff --name-only HEAD~3 -- '*.ts' '*.tsx' '*.js' '*.jsx' '*.py' '*.go' '*.rs' 2>/dev/null || \
git diff --name-only --cached -- '*.ts' '*.tsx' '*.js' '*.jsx' '*.py' '*.go' '*.rs' 2>/dev/null || \
git diff --name-only -- '*.ts' '*.tsx' '*.js' '*.jsx' '*.py' '*.go' '*.rs'

**Option B: User-specified scope**

If user mentions specific files or directories, use those instead.

**Option C: Recent conversation context**

Analyze conversation for files that were written/edited in this session.

Step 2: Load Simplification Configuration

Check for project-specific configuration:

if [ -f ".agent/.nav-config.json" ]; then
  python3 -c "import json; c=json.load(open('.agent/.nav-config.json')); s=c.get('simplification',{}); print('auto_apply=' + str(s.get('auto_apply', False)).lower()); print('enabled=' + str(s.get('enabled', True)).lower())"
fi

Default configuration:

{
  "simplification": {
    "enabled": true,
    "trigger": "post-implementation",
    "scope": "modified",
    "model": "opus",
    "skip_patterns": ["*.test.*", "*.spec.*", "*.md", "*.json", "*.yaml"],
    "max_file_size": 50000,
    "auto_apply": false,
    "preserve_comments": true
  }
}

Autonomous context check (CRITICAL)

After loading config, determine the invocation mode:

**Autonomous** (no human in the loop):

• Invoked from `nav-loop` VERIFY or COMPLETE phase
• Invoked from `nav-workflow` COMPLETE phase
• `NAV_AUTONOMOUS=1` env var set
• Triggered by autonomous-completion protocol

**Interactive** (human reviewing):

• Direct user request ("simplify this code")
• Mid-implementation manual invocation

**Decision matrix**:

| `auto_apply` | Context | Action | |---|---|---| | `true` | autonomous | Apply directly (Step 7 silent) | | `true` | interactive | Apply directly, show summary | | `false` | autonomous | **SKIP simplification entirely** — emit warning, return to caller. Never pause an autonomous flow waiting for approval. | | `false` | interactive | Show diff per file, prompt for approval |

When skipping in autonomous-mode-with-false, emit:

⚠️  nav-simplify skipped: auto_apply=false but invoked autonomously.
   To enable: set simplification.auto_apply=true in .agent/.nav-config.json
   To disable cleanly: set simplification.enabled=false

Step 3: Read Project Standards

**Load from CLAUDE.md** (required):

Read(file_path: "CLAUDE.md")

Extract coding standards:

• Preferred patterns (function vs arrow, explicit returns)
• Naming conventions
• Import ordering
• Error handling patterns
• Framework-specific guidelines

Step 4: Analyze Each File

For each modified file, analyze for simplification opportunities:

**Complexity Indicators** (look for):

• Deep nesting (> 3 levels)
• Long functions (> 50 lines)
• Redundant code patterns
• Unclear variable names
• Nested ternary operators
• Over-abstraction
• Dead code
• Inconsistent patterns

**Run analysis**:

PLUGIN_DIR="${CLAUDE_PLUGIN_ROOT:-$(cat "${NAVIGATOR_CONFIG_HOME:-${XDG_CONFIG_HOME:-$HOME/.config}/navigator}/plugin-root" 2>/dev/null)}"
[ -d "$PLUGIN_DIR/skills" ] || PLUGIN_DIR="$HOME/.claude/plugins/marketplaces/navigator-marketplace"
# Pass --scoring roi to enable cost/benefit gating (default: complexity,
# legacy behavior).
python3 "$PLUGIN_DIR/skills/nav-simplify/scripts/code_analyzer.py" --file "$file" --scoring roi

Step 4.5: ROI Gate (when scoring.mode = "roi")

If `simplification.scoring.mode` is `"roi"` in `.agent/.nav-config.json` (or `--scoring roi` was passed), the analyzer emits `benefit_score`, `cost_score`, `roi_score`, and a `gate_action` field. Honor the gate:

| `gate_action` | Behavior | |---|---| | `skip` | Skip this file. Emit a one-line reason with ROI math. Continue to next file. | | `suggest` | Force interactive mode for this file regardless of `auto_apply`. Show diff + reason. | | `apply` | Proceed with Step 5+ as normal (honor `auto_apply`). |

**Skip emission format**:

⏭️  {file} — low ROI (B={B}/10, C={C}/10, ROI={R}). Skipping.
   {reason}

`{reason}` summarizes the dominant cost factor (e.g., "stable file (412 days), 8 import references" or "below skip_below threshold").

**Mode is opt-in.** When `scoring.mode == "complexity"` (default), Step 4.5 is a no-op and behavior is unchanged from prior versions.

Step 5: Apply Simplification Rules

**Anthropic Simplification Rules**:

1. **Preserve Functionality**: Never change what the code does - only how it does it. All original features, outputs, and behaviors must re