#!/usr/bin/env python3
"""
Analyze Claude Code transcripts to extract real MCP vs non-MCP performance data
"""
import json
import os
from pathlib import Path
from typing import Dict, List, Any
import re
def analyze_transcript(file_path: Path) -> Dict[str, Any]:
"""Analyze a single Claude Code transcript."""
mcp_searches = []
traditional_searches = []
edits = []
with open(file_path, 'r') as f:
for line in f:
try:
entry = json.loads(line.strip())
# Check if this is an assistant message with tool uses
if entry.get('type') == 'assistant' and 'message' in entry:
message = entry['message']
# Look for tool uses in content
if isinstance(message.get('content'), list):
for content_item in message['content']:
if content_item.get('type') == 'tool_use':
tool = content_item.get('name', '')
tool_input = content_item.get('input', {})
# MCP search tools
if tool.startswith('mcp__') and ('search' in tool or 'lookup' in tool):
mcp_searches.append({
'tool': tool,
'query': tool_input.get('query') or tool_input.get('symbol', ''),
'timestamp': entry.get('timestamp', '')
})
# Traditional search tools
elif tool in ['Grep', 'Glob', 'Read', 'LS']:
traditional_searches.append({
'tool': tool,
'input': tool_input,
'timestamp': entry.get('timestamp', '')
})
# Edit tools
elif tool in ['Edit', 'MultiEdit', 'Write']:
edits.append({
'tool': tool,
'file': tool_input.get('file_path', ''),
'timestamp': entry.get('timestamp', '')
})
except json.JSONDecodeError:
continue
return {
'file': file_path.name,
'mcp_searches': mcp_searches,
'traditional_searches': traditional_searches,
'edits': edits,
'mcp_search_count': len(mcp_searches),
'traditional_search_count': len(traditional_searches),
'edit_count': len(edits)
}
def main():
"""Analyze all Claude Code transcripts."""
transcript_dir = Path.home() / '.claude' / 'projects' / '-workspaces-Code-Index-MCP'
if not transcript_dir.exists():
print(f"ā Transcript directory not found: {transcript_dir}")
return
transcripts = list(transcript_dir.glob('*.jsonl'))
print(f"š Found {len(transcripts)} Claude Code transcripts")
all_results = []
for transcript in transcripts:
print(f"\nš Analyzing {transcript.name}...")
result = analyze_transcript(transcript)
all_results.append(result)
print(f" MCP searches: {result['mcp_search_count']}")
print(f" Traditional searches: {result['traditional_search_count']}")
print(f" Edits: {result['edit_count']}")
# Show examples
if result['mcp_searches']:
print(f" Example MCP search: {result['mcp_searches'][0]['tool']} - '{result['mcp_searches'][0]['query']}'")
if result['traditional_searches']:
print(f" Example traditional: {result['traditional_searches'][0]['tool']}")
# Aggregate statistics
total_mcp = sum(r['mcp_search_count'] for r in all_results)
total_traditional = sum(r['traditional_search_count'] for r in all_results)
total_edits = sum(r['edit_count'] for r in all_results)
print("\n" + "="*60)
print("AGGREGATE STATISTICS")
print("="*60)
print(f"Total transcripts analyzed: {len(all_results)}")
print(f"Total MCP searches: {total_mcp}")
print(f"Total traditional searches: {total_traditional}")
print(f"Total edits: {total_edits}")
if total_mcp + total_traditional > 0:
mcp_ratio = total_mcp / (total_mcp + total_traditional) * 100
print(f"\nMCP search usage: {mcp_ratio:.1f}%")
print(f"Traditional search usage: {100 - mcp_ratio:.1f}%")
# Analyze edit patterns
edit_tools = {}
for result in all_results:
for edit in result['edits']:
tool = edit['tool']
edit_tools[tool] = edit_tools.get(tool, 0) + 1
if edit_tools:
print("\nEdit tool usage:")
for tool, count in sorted(edit_tools.items(), key=lambda x: x[1], reverse=True):
percentage = count / total_edits * 100
print(f" {tool}: {count} ({percentage:.1f}%)")
if tool in ['Edit', 'MultiEdit']:
print(f" ā Targeted edits (diffs)")
elif tool == 'Write':
print(f" ā Full file rewrites")
# Save detailed results
output = {
'summary': {
'transcripts_analyzed': len(all_results),
'total_mcp_searches': total_mcp,
'total_traditional_searches': total_traditional,
'total_edits': total_edits,
'mcp_search_ratio': total_mcp / max(1, total_mcp + total_traditional),
'edit_tool_distribution': edit_tools
},
'details': all_results
}
with open('claude_transcript_analysis.json', 'w') as f:
json.dump(output, f, indent=2)
print(f"\nš¾ Detailed results saved to claude_transcript_analysis.json")
if __name__ == "__main__":
main()
