extension/FEATURE_PARITY.md

Feature Parity Between Python MCP and TypeScript Extension

This document tracks the feature parity between the Python MCP implementation (~/work/hanzo/mcp) and the TypeScript extension (~/work/hanzo/extension).

Tool Parity Status

✅ Fully Implemented in Both

Tool Category	Tool Name	Python MCP	TypeScript Extension	Notes
File System	read	✅	✅
	write	✅	✅
	edit	✅	✅
	multi_edit	✅	✅
	directory_tree	✅	✅
	find_files	✅	✅
	grep	✅	✅
	git_search	✅	✅	Newly added
	content_replace	✅	✅	Newly added
	diff	✅	✅	Newly added
Search	symbols	✅	✅
	search	✅	✅	Unified search
	batch_search	✅	✅	Newly added
Shell	bash	✅	✅	Newly added
	run_command	✅	✅
	run_background	✅	✅	Newly added
	processes	✅	✅	Newly added
	pkill	✅	✅	Newly added
	logs	✅	✅	Newly added
	npx	✅	✅	Newly added
	uvx	✅	✅	Newly added
	open	✅	✅
Development	todo	✅	✅	Unified
	think	✅	✅
	critic	✅	✅	Newly added
AI/LLM	llm	✅	✅	Newly added
	consensus	✅	✅	Newly added
	agent	✅	✅	Newly added
	mode	✅	✅	Newly added
Database	sql_query	✅	✅
	sql_search	✅	✅
	sql_stats	✅	✅
	graph_*	✅	✅	Graph operations
Vector	vector_index	✅	✅
	vector_search	✅	✅
Utility	batch	✅	✅	Newly added - consolidated tool
	web_fetch	✅	✅
	rules	✅	✅
	config	✅	✅
System	stats	✅	✅
	tool_enable	✅	✅
	tool_disable	✅	✅
	tool_list	✅	✅
MCP	mcp	✅	✅	Newly added - manage arbitrary MCP servers

🚧 Python-Only Tools (Not Implemented in TypeScript)

Tool	Reason	Priority
watch	File watching in VS Code is limited	Low
neovim_*	Editor-specific tools	Low
mcp_*	MCP management tools	Medium

🚧 TypeScript-Only Tools

Tool	Reason	Priority
zen	VS Code specific implementation	N/A
palette	VS Code command palette integration	N/A

Test Coverage

New Test Files Added

1. batch-tools.test.ts

   • Sequential and parallel batch execution
   • Error handling and timeout support
   • Batch search with deduplication
   • Tool handler registration

2. ai-tools.test.ts

   • LLM provider integration (OpenAI, Anthropic, local)
   • Consensus mechanism across multiple LLMs
   • Agent delegation with different personalities
   • Development mode management

3. bash-tools.test.ts

   • Command execution with sessions
   • Background process management
   • Process monitoring and logging
   • NPX and UVX package runners

Key Improvements Made

1. Batch Tool Enhancement

   • Proper tool handler registration system
   • Support for parallel and sequential execution
   • Timeout handling per operation
   • Detailed execution results

2. AI Tools Integration

   • Unified LLM interface for multiple providers
   • Consensus mechanism for multi-LLM queries
   • Development mode system with 10+ personalities
   • Agent delegation for specialized tasks

3. Shell Tools Expansion

   • Full bash session support with persistence
   • Background process management
   • Process monitoring and log streaming
   • Package runner integration (npx, uvx)

4. Search Tools Enhancement

   • Git history and diff search
   • Content replacement across files
   • Batch search operations
   • File watching (limited implementation)

Authentication & Cloud Features

• ✅ Standalone authentication against iam.hanzo.ai
• ✅ Anonymous mode (--anon) for local-only usage
• ✅ OAuth flow with device ID generation
• ✅ Token refresh mechanism

Platform Support

• ✅ VS Code Extension
• ✅ Cursor IDE
• ✅ Windsurf IDE
• ✅ Claude Desktop
• ✅ Claude Code (via DXT)
• ✅ Standalone CLI via npx

Build System

All build targets are supported:

• npm run build - Standard VS Code extension
• npm run build:claude-desktop - Claude Desktop MCP server
• npm run build:dxt - Desktop Extension package
• npm run build:all - All targets

GitHub Actions automatically builds and releases all versions on tag push.

Next Steps

1. Performance Optimization

   • Implement caching for frequently used tools
   • Optimize batch operations for large datasets
   • Add streaming support for long-running operations

2. Enhanced Testing

   • Add integration tests for cross-tool workflows
   • Add performance benchmarks
   • Increase test coverage to >90%

3. Documentation

   • Complete API documentation for all tools
   • Add more usage examples
   • Create video tutorials

4. Feature Additions

   • Implement remaining Python-only tools if needed
   • Add more AI agent personalities
   • Enhance consensus mechanism with voting strategies