SSD-AI

AI Development Assistant based on Model Context Protocol

TypeScript + Python Support · 36 Specialized Tools · Intelligent Memory Management · Code Analysis · Reasoning Framework · Tasks Support

English | 한국어

---

Table of Contents

• Overview
• Key Features
• v1.6.0 Update
• Installation
• Tool Catalog
• Architecture
• Performance
• Development Guide
• License

---

Overview

Hi-AI is an AI development assistant that implements the Model Context Protocol (MCP) standard. It provides 36 specialized tools through natural language keyword recognition, helping developers perform complex tasks intuitively.

Core Values

• Natural Language: Execute tools automatically through Korean/English keywords
• Intelligent Memory: Context management and compression using SQLite
• Multi-Language Support: TypeScript, JavaScript, Python code analysis
• Performance Optimization: Project caching system
• Enterprise Quality: 100% test coverage and strict type system
• Long-Running Support: Task management for asynchronous operations
• Large-Scale Data: Cursor-based pagination

---

Key Features

1. Memory Management System

10 tools for maintaining context across sessions:

• Intelligent Storage: Information classification and priority management by category
• Context Compression: Priority-based context compression system
• Session Restoration: Perfect recreation of previous work states
• SQLite-Based: Concurrent control, indexing, transaction support

Key Tools:

• save_memory - Store information in long-term memory
• recall_memory - Search stored information
• auto_save_context - Automatic context saving
• restore_session_context - Session restoration
• prioritize_memory - Memory priority management

2. Semantic Code Analysis

AST-based code analysis and navigation tools:

• Symbol Search: Locate function, class, variable positions across projects
• Reference Tracking: Track all usages of specific symbols
• Multi-Language: TypeScript, JavaScript, Python support
• Project Caching: Performance optimization through LRU cache

Key Tools:

• find_symbol - Search for symbol definitions
• find_references - Find symbol references

3. Code Quality Analysis

Comprehensive code metrics and quality evaluation:

• Complexity Analysis: Cyclomatic, Cognitive, Halstead metrics
• Coupling/Cohesion: Structural soundness evaluation
• Quality Scores: A-F grade system
• Improvement Suggestions: Actionable refactoring recommendations

Key Tools:

• analyze_complexity - Complexity metric analysis
• validate_code_quality - Code quality evaluation
• check_coupling_cohesion - Coupling/cohesion analysis
• suggest_improvements - Improvement suggestions
• apply_quality_rules - Quality rule application
• get_coding_guide - Coding guide lookup

4. Project Planning Tools

Systematic requirements analysis and roadmap generation:

• PRD Generation: Automatic product requirements document creation
• User Stories: Story writing including acceptance criteria
• MoSCoW Analysis: Requirements prioritization
• Roadmap Creation: Step-by-step development schedule planning

Key Tools:

• generate_prd - Product requirements document generation
• create_user_stories - User story creation
• analyze_requirements - Requirements analysis
• feature_roadmap - Feature roadmap creation

5. Sequential Thinking Tools

Structured problem solving and decision making support:

• Problem Decomposition: Break down complex problems step by step
• Thinking Chains: Sequential reasoning process generation
• Multiple Perspectives: Analytical/Creative/Systematic/Critical thinking
• Execution Plans: Convert tasks into executable plans

Key Tools:

• create_thinking_chain - Thinking chain creation
• analyze_problem - Problem analysis
• step_by_step_analysis - Step-by-step analysis
• break_down_problem - Problem decomposition
• think_aloud_process - Thinking process expression
• format_as_plan - Plan formatting

6. Prompt Engineering

Prompt quality improvement and optimization:

• Automatic Enhancement: Convert vague requests to specific ones
• Quality Evaluation: Score clarity, specificity, contextuality
• Structuring: Goal, background, requirements, quality criteria

Key Tools:

• enhance_prompt - Prompt enhancement
• analyze_prompt - Prompt quality analysis

7. Browser Automation

Web-based debugging and testing:

• Console Monitoring: Browser console log capture
• Network Analysis: HTTP request/response tracking
• Cross-Platform: Chrome, Edge, Brave support

Key Tools:

• monitor_console_logs - Console log monitoring
• inspect_network_requests - Network request analysis

8. UI Preview

Pre-coding UI layout visualization:

• ASCII Art: Support for 6 layout types
• Responsive Preview: Desktop/mobile views
• Pre-Approval: Confirm structure before coding

Key Tools:

• preview_ui_ascii - ASCII UI preview

9. Time Utilities

Various format time queries:

Key Tools:

• get_current_time - Current time query (ISO, UTC, timezones, etc.)

10. Tasks and Pagination Support

Long-running operations and large-scale data processing:

• Tasks: MCP 2025-11-25 experimental feature for long-running task management
• Pagination: Cursor-based pagination for large dataset processing
• Asynchronous Operations: Execute complex analysis tasks in background
• Status Tracking: Real-time task progress monitoring

Tasks-Enabled Tools:

• find_symbol, find_references (semantic analysis)
• analyze_complexity, check_coupling_cohesion, validate_code_quality, suggest_improvements (code quality)
• analyze_requirements, feature_roadmap, generate_prd (project planning)
• apply_reasoning_framework, enhance_prompt_gemini (reasoning and prompts)

---

v1.6.0 Update

New Features (2025-01-27)

Tasks Support (Experimental MCP Feature)

Long-Running Task Management

• Implementation of MCP 2025-11-25 Tasks specification
• Execute complex analysis tasks in background
• Real-time task status tracking and monitoring
• TTL-based automatic cleanup (default 5 minutes, max 1 hour)

Tasks API

• tasks/get - Query task status
• tasks/result - Query task result (wait until completion)
• tasks/list - List all tasks (with pagination)
• tasks/cancel - Cancel running task
• notifications/tasks/status - Status change notifications

Task-Enabled Tools (11 tools)

• Semantic Analysis: find_symbol, find_references
• Code Quality: analyze_complexity, check_coupling_cohesion, validate_code_quality, suggest_improvements
• Project Planning: analyze_requirements, feature_roadmap, generate_prd
• Reasoning/Prompts: apply_reasoning_framework, enhance_prompt_gemini

Pagination Support

Cursor-Based Pagination

• MCP specification compliant cursor-based implementation
• Efficient processing of large lists
• Enhanced security through opaque cursors

Supported List Operations

• tools/list - Tool list (20 items by default)
• resources/list - Resource list
• prompts/list - Prompt list
• tasks/list - Task list

Integration Effects

• Asynchronous Operation Support: Execute complex analysis in background
• Large-Scale Data Processing: Improved memory efficiency through pagination
• Real-Time Monitoring: Task progress tracking
• Enhanced User Experience: Perform other tasks during long operations

---

Installation

System Requirements

• Node.js 18.0 or higher
• TypeScript 5.0 or higher
• MCP-compatible client (Claude Desktop, Cursor, Windsurf)
• Python 3.x (for Python code analysis)

Installation Methods

NPM Package

# Global installation
npm install -g @ssdeanx/ssd-ai

# Local installation
npm install @ssdeanx/ssd-ai

Smithery Platform

# One-click installation
https://smithery.ai/server/@su-record/hi-ai

MCP Client Configuration

Add to your Claude Desktop or other MCP client's configuration file:

{
  "mcpServers": {
    "hi-ai": {
      "command": "hi-ai",
      "args": [],
      "env": {}
    }
  }
}

---

Tool Catalog

Complete Tool List (36 tools)

Category	Count	Tool List
Memory	10	save_memory, recall_memory, list_memories, search_memories, delete_memory, update_memory, auto_save_context, restore_session_context, prioritize_memory, start_session
Semantic	2	find_symbol, find_references
Thinking	6	create_thinking_chain, analyze_problem, step_by_step_analysis, break_down_problem, think_aloud_process, format_as_plan
Reasoning	1	apply_reasoning_framework
Code Quality	6	analyze_complexity, validate_code_quality, check_coupling_cohesion, suggest_improvements, apply_quality_rules, get_coding_guide
Planning	4	generate_prd, create_user_stories, analyze_requirements, feature_roadmap
Prompt	2	enhance_prompt, analyze_prompt
Browser	2	monitor_console_logs, inspect_network_requests
UI	1	preview_ui_ascii
Time	1	get_current_time

Tasks-Enabled Tools (11 tools)

The following tools support long-running operations through Tasks:

• Semantic Analysis: find_symbol, find_references
• Code Quality: analyze_complexity, check_coupling_cohesion, validate_code_quality, suggest_improvements
• Project Planning: analyze_requirements, feature_roadmap, generate_prd
• Reasoning/Prompts: apply_reasoning_framework, enhance_prompt_gemini

Keyword Mapping Examples

Memory Tools

Tool	English	Korean
save_memory	remember, save this	기억해, 저장해
recall_memory	recall, remind me	떠올려, 기억나
auto_save_context	commit, checkpoint	커밋, 저장

Code Analysis Tools

Tool	English	Korean
find_symbol	find function, where is	함수 찾아, 클래스 어디
analyze_complexity	complexity, how complex	복잡도, 복잡한지
validate_code_quality	quality, review	품질, 리뷰

Tasks Tools

Tool	English	Korean
tasks/get	task status, progress	작업 상태, 진행 상황
tasks/result	get result, wait for completion	결과 가져와, 완료될 때까지
tasks/cancel	cancel task, stop	작업 취소, 중지해

---

Architecture

System Structure

graph TB
    subgraph "Client Layer"
        A[Claude Desktop / Cursor / Windsurf]
    end

    subgraph "MCP Server"
        B[Hi-AI v1.6.0]
    end

    subgraph "Core Libraries"
        C1[MemoryManager]
        C2[ContextCompressor]
        C3[ProjectCache]
        C4[PythonParser]
        C5[TaskManager]
    end

    subgraph "Tool Categories"
        D1[Memory Tools x10]
        D2[Semantic Tools x2]
        D3[Thinking Tools x6]
        D4[Quality Tools x6]
        D5[Planning Tools x4]
        D6[Prompt Tools x2]
        D7[Browser Tools x2]
        D8[UI Tools x1]
        D9[Time Tools x1]
        D10[Tasks Support]
    end

    subgraph "Data Layer"
        E1[(SQLite Database)]
        E2[Project Files]
        E3[Task Store]
    end

    A <--> B
    B --> C1 & C2 & C3 & C4 & C5
    B --> D1 & D2 & D3 & D4 & D5 & D6 & D7 & D8 & D9 & D10
    C1 --> E1
    C3 --> E2
    C4 --> E2
    C5 --> E3
    D1 --> C1 & C2
    D2 --> C3 & C4
    D4 --> C4
    D10 --> C5

Core Components

TaskManager

• Role: Lifecycle management of long-running tasks
• Features: Task creation, status tracking, result storage, TTL management
• States: working, input_required, completed, failed, cancelled
• Notifications: Real-time status change notifications

Pagination System

• Role: Efficient processing of large list data
• Method: Cursor-based pagination
• Security: Prevent data exposure through opaque cursors

Data Flow

User Input (Natural Language)
    ↓
Keyword Matching (Tool Selection)
    ↓
Tasks Support Check
    ↓
Normal Execution or Task Creation
    ↓
Asynchronous Execution (Tasks)
    ↓
Status Polling or Real-time Notifications
    ↓
Result Return

---

Performance

Major Optimizations

Project Caching

• Performance improvement for repeated analysis through LRU cache
• Maintain latest state with 5-minute TTL
• Resource management through memory limits

Memory Operations

• Batch operation optimization through SQLite transactions
• Time complexity improvement: O(n²) → O(n)
• Fast lookup through indexing

Tasks Optimization

• Improved UI responsiveness through background execution
• Prevent memory leaks through TTL-based automatic cleanup
• Efficient monitoring through status-based polling

Response Format

• Switch to concise response format
• Output focused on core information

v1.5.0 Response Example:

{
  "action": "save_memory",
  "key": "test-key",
  "value": "test-value",
  "category": "general",
  "timestamp": "2025-01-16T12:34:56.789Z",
  "status": "success",
  "metadata": { ... }
}

v1.6.0 Response Example:

✓ Saved: test-key
Category: general

---

Development Guide

Environment Setup

# Clone repository
git clone https://github.com/ssdeanx/ssd-ai.git
cd ssd-ai

# Install dependencies
npm install

# Build
npm run build

# Development mode
npm run dev

Testing

# Run all tests
npm test

# Watch mode
npm run test:watch

# UI mode
npm run test:ui

# Coverage report
npm run test:coverage

Code Style

• TypeScript: strict mode
• Types: Use src/types/tool.ts
• Tests: Maintain 100% coverage
• Commits: Conventional Commits format

Adding New Tools

1. Create file in src/tools/category/ directory
2. Implement ToolDefinition interface
3. Register tool in src/index.ts
4. Write tests in tests/unit/ directory
5. Update README

Pull Request

1. Create feature branch: feature/tool-name
2. Write and pass tests
3. Confirm successful build
4. Create PR and request review

---

Contributors

Special Thanks

• Smithery - MCP server deployment and one-click installation platform

---

License

MIT License - Free to use, modify, and distribute

---

Citation

If you use this project for research or commercial purposes:

@software{hi-ai2024,
  author = {ssdeanx},
  title = {Hi-AI: Natural Language MCP Server for AI-Assisted Development},
  year = {2024},
  version = {1.6.0},
  url = {https://github.com/su-record/hi-ai}
}

---

Star History

Hi-AI v1.6.0

Tasks Support · Cursor-Based Pagination · 36 Specialized Tools · 122 Tests · 100% Coverage

Made with ❤️ by Su

🏠 Homepage · 📚 Documentation · 🐛 Issues · 💬 Discussions