Qwen CLI Refactoring Skill

Agent: Qwen 1.5B (strategic analysis + code extraction) Validation: Gemma 270M (pattern fidelity check) Token Budget: 1,300 tokens (800 extraction + 400 refactoring + 100 validation)

Skill Purpose

Refactor monolithic CLI files (>1,000 lines) by extracting logical command modules while preserving all functionality. Uses Qwen for strategic analysis and module extraction, with Gemma validation for pattern fidelity.

Trigger Source: Manual invocation by 0102 when CLI files exceed WSP 49 limits

Success Criteria:

• Reduce main() function size by >70%
• Extract 5+ independent command modules
• Zero regressions (all flags work identically)
• Pattern fidelity >90% (Gemma validation)

Input Context

{
  "file_path": "path/to/cli.py",
  "current_lines": 1470,
  "main_function_lines": 1144,
  "target_reduction_percent": 70,
  "preserve_flags": ["--search", "--index", "--all-67-flags"],
  "output_directory": "path/to/cli/commands/"
}

Micro Chain-of-Thought Steps

Step 1: Analyze CLI Structure (200 tokens)

Qwen Analysis Task: Read cli.py and identify:

1. Command-line argument groups (search, index, holodae, etc.)
2. Logical sections in main() function
3. Shared dependencies between sections
4. Natural module boundaries

Output:

{
  "total_lines": 1470,
  "main_function_lines": 1144,
  "argument_groups": [
    {"name": "search", "flags": ["--search", "--limit"], "lines": [601, 750]},
    {"name": "index", "flags": ["--index-all", "--index-code"], "lines": [751, 900]},
    {"name": "holodae", "flags": ["--start-holodae", "--stop-holodae"], "lines": [901, 1050]},
    {"name": "module", "flags": ["--link-modules", "--query-modules"], "lines": [1051, 1200]},
    {"name": "codeindex", "flags": ["--code-index-report"], "lines": [1201, 1350]}
  ],
  "shared_dependencies": ["throttler", "reward_events", "args"],
  "extraction_priority": ["search", "index", "holodae", "module", "codeindex"]
}

Step 2: Extract Command Modules (400 tokens)

Qwen Extraction Task: For each command group:

1. Extract code from main() function
2. Create commands/{name}.py file
3. Convert to class-based command pattern
4. Preserve all flag handling logic

Template Pattern:

# commands/search.py
from typing import Any, Dict
from ..core import HoloIndex

class SearchCommand:
    def __init__(self, holo_index: HoloIndex):
        self.holo_index = holo_index

    def execute(self, args, throttler, add_reward_event) -> Dict[str, Any]:
        \"\"\"Execute search command with preserved flag logic\"\"\"
        # [EXTRACTED CODE FROM MAIN() LINES 601-750]
        results = self.holo_index.search(args.search, limit=args.limit)
        return {"results": results, "success": True}

Output: 5 new command module files created

Step 3: Refactor main() Function (200 tokens)

Qwen Refactoring Task:

1. Remove extracted code from main()
2. Add command routing logic
3. Instantiate command classes
4. Delegate execution to appropriate command

New main() Structure:

def main() -> None:
    args = parser.parse_args()
    throttler = AgenticOutputThrottler()

    # Initialize HoloIndex
    holo_index = HoloIndex(...)

    # Command routing
    if args.search:
        from .commands.search import SearchCommand
        cmd = SearchCommand(holo_index)
        result = cmd.execute(args, throttler, add_reward_event)
    elif args.index or args.index_all:
        from .commands.index import IndexCommand
        cmd = IndexCommand(holo_index)
        result = cmd.execute(args, throttler, add_reward_event)
    # ... etc for other commands

    # Render output (preserved logic)
    render_response(throttler, result, args)

Output: Refactored main.py (reduced from 1,144 → ~300 lines)

Step 4: Gemma Pattern Fidelity Validation (100 tokens)

Gemma Validation Task: Compare original vs refactored:

1. All 67 flags still recognized
2. Execution flow unchanged
3. Output format identical
4. No missing imports

Validation Checks:

original_flags = extract_flags("cli.py")
refactored_flags = extract_flags("cli/main.py") + extract_flags("cli/commands/*.py")

assert set(original_flags) == set(refactored_flags), "Missing flags detected"
assert pattern_fidelity >= 0.90, "Pattern fidelity below threshold"

Output:

{
  "pattern_fidelity": 0.95,
  "flags_preserved": 67,
  "missing_flags": [],
  "regressions_detected": 0,
  "validation_passed": true
}

Step 5: Generate Migration Report (100 tokens)

Report Contents:

1. Files created (5 command modules)
2. main() reduction (1,144 → 300 lines, 74% reduction)
3. Validation results (fidelity: 95%)
4. Token cost (actual vs estimated)
5. Next steps (testing, documentation)

Output:

# CLI Refactoring Report

**Date**: 2025-10-25
**File**: holo_index/cli.py
**Status**: COMPLETE ✅

## Changes Summary
- main() reduced: 1,144 → 300 lines (74% reduction)
- Command modules created: 5
- Total lines: 1,470 → 1,350 (distributed across 6 files)
- Pattern fidelity: 95% (Gemma validated)

## Files Created
1. cli/commands/search.py (200 lines)
2. cli/commands/index.py (180 lines)
3. cli/commands/holodae.py (190 lines)
4. cli/commands/module.py (210 lines)
5. cli/commands/codeindex.py (170 lines)

## Validation
- ✅ All 67 flags preserved
- ✅ Zero regressions detected
- ✅ Pattern fidelity: 95%
- ✅ Imports resolved

## Token Cost
- Estimated: 1,300 tokens
- Actual: 1,150 tokens (12% under budget)

## Next Steps
1. Run integration tests
2. Update documentation
3. Commit with 0102 approval

Execution Constraints

Authorized Actions (Autonomous)

• ✅ Create new files in cli/commands/ directory
• ✅ Extract code from main() function
• ✅ Update imports in main.py
• ✅ Run Gemma validation checks

Requires 0102 Approval

• ❌ Modifying flag names
• ❌ Removing any flags
• ❌ Changing command behavior
• ❌ Committing changes to git

Safety Guardrails

1. Backup: Create cli.py.backup before modification
2. Validation: Gemma fidelity must be ≥90%
3. Rollback: Restore backup if validation fails
4. Reporting: Report progress after each extraction

Pattern Memory Storage

After successful execution, store refactoring pattern:

{
  "pattern_name": "cli_refactoring",
  "original_size": 1470,
  "refactored_size": 1350,
  "main_reduction": 0.74,
  "modules_extracted": 5,
  "token_cost": 1150,
  "fidelity": 0.95,
  "success": true,
  "learned": "Extract commands by flag groups, preserve shared state via dependency injection"
}

Example Invocation

Via WRE Master Orchestrator:

from modules.infrastructure.wre_core.wre_master_orchestrator import WREMasterOrchestrator

orchestrator = WREMasterOrchestrator()

result = orchestrator.execute_skill(
    skill_name="qwen_cli_refactor",
    agent="qwen",
    input_context={
        "file_path": "holo_index/cli.py",
        "current_lines": 1470,
        "main_function_lines": 1144,
        "target_reduction_percent": 70,
        "output_directory": "holo_index/cli/commands/"
    }
)

print(f"Refactoring {'succeeded' if result['success'] else 'failed'}")
print(f"Pattern fidelity: {result['pattern_fidelity']}")
print(f"Token cost: {result['token_cost']}")

WSP Compliance

References:

• WSP 49: Module Structure (file size limits)
• WSP 72: Block Independence (command isolation)
• WSP 50: Pre-Action Verification (backup before modification)
• WSP 96: WRE Skills Protocol (this skill definition)

Success Metrics

Next Evolution: After 10+ successful executions, promote from prototype → production