@rfxlamia/skillkit 1.0.0

package/skills/skillkit/scripts/split_skill.py

@@ -0,0 +1,540 @@
+#!/usr/bin/env python3
+"""
+Auto-split large SKILL.md for progressive disclosure optimization.
+Based on token economics from File 05 and architecture from File 10.
+
+References: Files 05, 10 (token optimization, architecture)
+"""
+
+import re
+import json
+import argparse
+import sys
+from pathlib import Path
+from typing import List, Dict, Optional
+
+# Import shared utilities for standardized output
+try:
+    from utils.output_formatter import add_format_argument, format_success_response, format_error_response, output_json
+except ImportError:
+    # Fallback if utils not in path
+    sys.path.insert(0, str(Path(__file__).parent))
+    from utils.output_formatter import add_format_argument, format_success_response, format_error_response, output_json
+
+class SkillSplitter:
+    """Intelligently split SKILL.md into main + references."""
+    
+    def __init__(self, 
+                 skill_path: str, 
+                 threshold: int = 500, 
+                 preview: bool = False,
+                 auto: bool = False,
+                 output_format: str = 'text'):
+        """
+        Initialize splitter.
+        
+        Args:
+            skill_path: Path to skill directory
+            threshold: Line count threshold for splitting
+            preview: If True, show changes without applying
+            auto: If True, split without confirmation
+            output_format: 'text' or 'json' for agent-layer
+        
+        References: File 10 (progressive disclosure architecture)
+        """
+        self.skill_path = Path(skill_path)
+        self.threshold = threshold
+        self.preview = preview
+        self.auto = auto
+        self.output_format = output_format
+        self.skill_md_content = None
+        self.frontmatter = ''
+        self.sections = []
+        self.core_sections = []
+        self.reference_sections = []
+    
+    # ========== ANALYSIS ==========
+    
+    def analyze_skill(self) -> Dict:
+        """
+        Analyze SKILL.md structure and identify split candidates.
+        
+        Returns:
+            Dict with analysis results including:
+            - needs_split: bool
+            - total_lines: int
+            - sections: int
+            - core_sections: int
+            - reference_sections: int
+        
+        Raises:
+            FileNotFoundError: If SKILL.md doesn't exist
+        
+        References: File 05 (token optimization strategies)
+        """
+        skill_md = self.skill_path / "SKILL.md"
+        if not skill_md.exists():
+            raise FileNotFoundError(f"SKILL.md not found in {self.skill_path}")
+        
+        with open(skill_md, encoding='utf-8') as f:
+            self.skill_md_content = f.read()
+            lines = self.skill_md_content.split('\n')
+        
+        total_lines = len(lines)
+        
+        if self.output_format == 'text':
+            print(f"\nAnalyzing SKILL.md...")
+            print(f"  Current size: {total_lines} lines")
+            print(f"  Threshold: {self.threshold} lines")
+        
+        if total_lines < self.threshold:
+            if self.output_format == 'text':
+                print(f"\n✔ SKILL.md already optimal (<{self.threshold} lines)")
+                print("   No split needed.")
+            return {
+                'needs_split': False,
+                'total_lines': total_lines,
+                'threshold': self.threshold
+            }
+        
+        if self.output_format == 'text':
+            print(f"\n⚠️  SKILL.md exceeds threshold ({total_lines} > {self.threshold})")
+        
+        # Parse structure
+        self._parse_structure()
+        self._classify_sections()
+        
+        core_lines = sum(s['lines'] for s in self.core_sections)
+        ref_lines = sum(s['lines'] for s in self.reference_sections)
+        
+        return {
+            'needs_split': True,
+            'total_lines': total_lines,
+            'threshold': self.threshold,
+            'core_sections': len(self.core_sections),
+            'reference_sections': len(self.reference_sections),
+            'core_lines': core_lines,
+            'reference_lines': ref_lines,
+            'reduction_percent': (ref_lines / total_lines * 100) if total_lines > 0 else 0
+        }
+    
+    def _parse_structure(self):
+        """Parse SKILL.md into sections."""
+        lines = self.skill_md_content.split('\n')
+        
+        # Extract frontmatter
+        in_frontmatter = False
+        frontmatter_lines = []
+        content_start = 0
+        
+        for i, line in enumerate(lines):
+            if line.strip() == '---':
+                if not in_frontmatter:
+                    in_frontmatter = True
+                    frontmatter_lines.append(line)
+                elif in_frontmatter:
+                    frontmatter_lines.append(line)
+                    self.frontmatter = '\n'.join(frontmatter_lines)
+                    content_start = i + 1
+                    break
+            elif in_frontmatter:
+                frontmatter_lines.append(line)
+        
+        # Parse sections (## headers)
+        current_section = None
+        
+        for i in range(content_start, len(lines)):
+            line = lines[i]
+            
+            # Check for ## header
+            if line.startswith('## '):
+                # Save previous section
+                if current_section:
+                    self.sections.append(current_section)
+                
+                # Start new section
+                current_section = {
+                    'header': line,
+                    'title': line.strip('# ').strip(),
+                    'content': '',
+                    'lines': 0,
+                    'type': 'unknown'
+                }
+            elif current_section:
+                current_section['content'] += line + '\n'
+                current_section['lines'] += 1
+        
+        # Save last section
+        if current_section:
+            self.sections.append(current_section)
+    
+    def _classify_sections(self):
+        """Classify sections as core or reference."""
+        # Core section patterns
+        core_patterns = [
+            r'overview',
+            r'quick start',
+            r'installation',
+            r'usage',
+            r'basic',
+            r'getting started',
+            r'introduction'
+        ]
+        
+        # Reference section patterns
+        reference_patterns = [
+            r'advanced',
+            r'detailed',
+            r'examples',
+            r'reference',
+            r'appendix',
+            r'troubleshooting',
+            r'faq',
+            r'api',
+            r'configuration'
+        ]
+        
+        for section in self.sections:
+            title_lower = section['title'].lower()
+            
+            # Check if core
+            if any(re.search(pattern, title_lower) for pattern in core_patterns):
+                section['type'] = 'core'
+                self.core_sections.append(section)
+            # Check if reference
+            elif any(re.search(pattern, title_lower) for pattern in reference_patterns):
+                section['type'] = 'reference'
+                self.reference_sections.append(section)
+            # Default: keep in core if short, move to reference if long
+            elif section['lines'] < 50:
+                section['type'] = 'core'
+                self.core_sections.append(section)
+            else:
+                section['type'] = 'reference'
+                self.reference_sections.append(section)
+    
+    # ========== SPLITTING ==========
+    
+    def perform_split(self) -> Dict:
+        """
+        Perform the actual split operation.
+        
+        Returns:
+            Dict with split results
+        """
+        if not self.sections:
+            raise ValueError("No sections found. Run analyze_skill() first.")
+        
+        core_lines = sum(s['lines'] for s in self.core_sections)
+        ref_lines = sum(s['lines'] for s in self.reference_sections)
+        total_lines = core_lines + ref_lines
+        
+        # Show preview if requested
+        if self.preview:
+            if self.output_format == 'text':
+                self._show_split_preview()
+            return {
+                'executed': False,
+                'preview_mode': True,
+                'core_lines': core_lines,
+                'reference_lines': ref_lines,
+                'total_lines': total_lines,
+                'core_sections': [{'title': s['title'], 'lines': s['lines']} for s in self.core_sections],
+                'reference_sections': [{'title': s['title'], 'lines': s['lines']} for s in self.reference_sections],
+                'reduction_percent': (ref_lines / total_lines * 100) if total_lines > 0 else 0
+            }
+        
+        # Confirm with user (unless auto mode or JSON output)
+        if not self.auto and self.output_format == 'text':
+            print()
+            response = input("Proceed with split? [y/n]: ")
+            if response.lower() != 'y':
+                print("Split cancelled.")
+                return {
+                    'executed': False,
+                    'cancelled': True
+                }
+        
+        # Execute split
+        if self.output_format == 'text':
+            print("\nExecuting split...")
+        
+        # Create references directory
+        refs_dir = self.skill_path / "references"
+        refs_dir.mkdir(exist_ok=True)
+        if self.output_format == 'text':
+            print(f"  • Created directory: references/")
+        
+        # Move reference sections
+        created_files = []
+        for section in self.reference_sections:
+            filename = self._create_reference_file(section, refs_dir)
+            created_files.append(filename)
+        
+        # Update SKILL.md
+        self._update_skill_md()
+        
+        # Generate report
+        self._generate_split_report(core_lines, ref_lines, total_lines)
+        
+        if self.output_format == 'text':
+            print(f"\n✔ Split completed successfully!")
+        
+        return {
+            'executed': True,
+            'core_lines': core_lines,
+            'reference_lines': ref_lines,
+            'total_lines': total_lines,
+            'files_created': created_files,
+            'core_sections': [{'title': s['title'], 'lines': s['lines']} for s in self.core_sections],
+            'reference_sections': [{'title': s['title'], 'lines': s['lines'], 'file': s.get('reference_file', '')} for s in self.reference_sections],
+            'reduction_percent': (ref_lines / total_lines * 100) if total_lines > 0 else 0
+        }
+    
+    def _create_reference_file(self, section: Dict, refs_dir: Path) -> str:
+        """
+        Create a reference file from a section.
+        
+        Args:
+            section: Section dictionary
+            refs_dir: References directory path
+        
+        Returns:
+            Filename created
+        """
+        # Generate filename from title
+        filename = self._sanitize_filename(section['title']) + '.md'
+        filepath = refs_dir / filename
+        
+        # Write content
+        with open(filepath, 'w', encoding='utf-8') as f:
+            f.write(f"# {section['title']}\n\n")
+            f.write(section['content'].strip())
+            f.write('\n')  # Ensure trailing newline
+        
+        if self.output_format == 'text':
+            print(f"  • Created: references/{filename} ({section['lines']} lines)")
+        
+        # Store reference for updating SKILL.md
+        section['reference_file'] = filename
+        
+        return filename
+    
+    def _sanitize_filename(self, title: str) -> str:
+        """
+        Convert section title to valid filename.
+        
+        Args:
+            title: Section title
+        
+        Returns:
+            Sanitized filename (lowercase, hyphens, no special chars)
+        """
+        # Remove special chars, convert spaces to hyphens
+        filename = re.sub(r'[^a-zA-Z0-9\s-]', '', title)
+        filename = filename.lower().replace(' ', '-')
+        # Remove multiple consecutive hyphens
+        filename = re.sub(r'-+', '-', filename)
+        # Remove leading/trailing hyphens
+        filename = filename.strip('-')
+        return filename
+    
+    def _update_skill_md(self):
+        """
+        Update SKILL.md with core sections and cross-references.
+        
+        New structure:
+        1. Frontmatter (preserved)
+        2. Core sections (preserved)
+        3. Additional Resources section (new, with links)
+        """
+        # Build new SKILL.md
+        new_content = []
+        
+        # Add frontmatter
+        if self.frontmatter:
+            new_content.append(self.frontmatter.rstrip())
+        
+        # Add core sections
+        for section in self.core_sections:
+            new_content.append(f"\n{section['header']}")
+            new_content.append(section['content'].strip())
+        
+        # Add reference links section
+        new_content.append("\n## Additional Resources")
+        new_content.append("\nFor detailed information, see:")
+        new_content.append("")
+        
+        for section in self.reference_sections:
+            ref_file = section['reference_file']
+            new_content.append(f"- [{section['title']}](references/{ref_file})")
+        
+        # Write updated SKILL.md
+        skill_md = self.skill_path / "SKILL.md"
+        with open(skill_md, 'w', encoding='utf-8') as f:
+            f.write('\n'.join(new_content))
+            f.write('\n')  # Ensure trailing newline
+        
+        if self.output_format == 'text':
+            print(f"  • Updated: SKILL.md")
+    
+    def _show_split_preview(self):
+        """Show preview of split without applying changes."""
+        print("\n" + "="*70)
+        print("SPLIT PREVIEW")
+        print("="*70)
+        
+        print("\n📄 Core Sections (will remain in SKILL.md):")
+        for i, section in enumerate(self.core_sections, 1):
+            print(f"  {i}. {section['title']} ({section['lines']} lines)")
+        
+        print(f"\n📚 Reference Sections (will move to references/):")
+        for i, section in enumerate(self.reference_sections, 1):
+            filename = self._sanitize_filename(section['title']) + '.md'
+            print(f"  {i}. {section['title']}")
+            print(f"     → references/{filename} ({section['lines']} lines)")
+        
+        print("\n" + "="*70)
+        core_lines = sum(s['lines'] for s in self.core_sections)
+        ref_lines = sum(s['lines'] for s in self.reference_sections)
+        total_lines = core_lines + ref_lines
+        
+        print(f"Summary:")
+        print(f"  Current SKILL.md: {total_lines} lines")
+        print(f"  After split: {core_lines} lines (SKILL.md)")
+        print(f"  References: {len(self.reference_sections)} files, {ref_lines} lines total")
+        print(f"  Reduction: {ref_lines/total_lines*100:.1f}%")
+        print("="*70)
+    
+    def _generate_split_report(self, core_lines: int, ref_lines: int, total_lines: int):
+        """
+        Generate and save split report.
+        
+        Args:
+            core_lines: Lines remaining in SKILL.md
+            ref_lines: Lines moved to references
+            total_lines: Original total lines
+        """
+        report_path = self.skill_path / "split_report.md"
+        
+        with open(report_path, 'w', encoding='utf-8') as f:
+            f.write("# Skill Split Report\n\n")
+            f.write(f"**Date:** {self._get_date()}\n")
+            f.write(f"**Threshold:** {self.threshold} lines\n\n")
+            
+            f.write("## Summary\n\n")
+            f.write(f"- **Original size:** {total_lines} lines\n")
+            f.write(f"- **New SKILL.md size:** {core_lines} lines\n")
+            f.write(f"- **References created:** {len(self.reference_sections)} files\n")
+            f.write(f"- **Total reference lines:** {ref_lines} lines\n")
+            f.write(f"- **Size reduction:** {ref_lines/total_lines*100:.1f}%\n\n")
+            
+            f.write("## Core Sections (in SKILL.md)\n\n")
+            for section in self.core_sections:
+                f.write(f"- {section['title']} ({section['lines']} lines)\n")
+            
+            f.write("\n## Reference Sections (in references/)\n\n")
+            for section in self.reference_sections:
+                f.write(f"- [{section['title']}](references/{section['reference_file']}) ")
+                f.write(f"({section['lines']} lines)\n")
+        
+        if self.output_format == 'text':
+            print(f"  • Generated: split_report.md")
+    
+    def _get_date(self) -> str:
+        """Get current date in YYYY-MM-DD format."""
+        from datetime import datetime
+        return datetime.now().strftime("%Y-%m-%d")
+
+def main():
+    """CLI entry point."""
+    parser = argparse.ArgumentParser(
+        description="Auto-split large SKILL.md for progressive disclosure",
+        epilog="References: Files 05, 10 for optimization strategies"
+    )
+    parser.add_argument(
+        'skill_path',
+        type=str,
+        help='Path to skill directory'
+    )
+    parser.add_argument(
+        '--threshold',
+        type=int,
+        default=500,
+        help='Line count threshold for splitting (default: 500)'
+    )
+    parser.add_argument(
+        '--preview',
+        action='store_true',
+        help='Preview changes without applying'
+    )
+    parser.add_argument(
+        '--auto',
+        action='store_true',
+        help='Automatic split without confirmation'
+    )
+    add_format_argument(parser)  # Standardized --format argument
+
+    args = parser.parse_args()
+    
+    try:
+        splitter = SkillSplitter(
+            args.skill_path,
+            threshold=args.threshold,
+            preview=args.preview,
+            auto=args.auto,
+            output_format=args.format
+        )
+
+        analysis = splitter.analyze_skill()
+
+        if args.format == 'json':
+            # JSON output mode
+            if analysis.get('needs_split'):
+                result = splitter.perform_split()
+                output = {
+                    'status': 'success',
+                    'analysis': analysis,
+                    'split_result': result
+                }
+            else:
+                output = {
+                    'status': 'success',
+                    'analysis': analysis,
+                    'message': 'No split needed - skill already optimal'
+                }
+            print(json.dumps(output, indent=2))
+        else:
+            # Text output mode (existing behavior)
+            if analysis.get('needs_split'):
+                splitter.perform_split()
+        
+        return 0
+    
+    except FileNotFoundError as e:
+        if args.format == 'json':
+            response = format_error_response(
+                error_type='FileNotFoundError',
+                message=str(e),
+                tool_name='split_skill',
+                help_text='Ensure skill directory exists and contains SKILL.md'
+            )
+            output_json(response)
+        else:
+            print(f"\n✗ Error: {e}")
+        return 1
+    except Exception as e:
+        if args.format == 'json':
+            response = format_error_response(
+                error_type=type(e).__name__,
+                message=str(e),
+                tool_name='split_skill',
+                help_text='Check skill structure and permissions'
+            )
+            output_json(response)
+        else:
+            print(f"\n✗ Unexpected error: {e}")
+        return 2
+
+if __name__ == "__main__":
+    exit(main())