Hi-AI

Hi-AI is an MCP-compliant AI development assistant that uses natural-language keywords to activate a rich set of 34 specialized tools. It maintains a knowledge-graph memory, supports multi-language code analysis, and emphasizes architecture-aware tooling to help you design, analyze, and optimize AI-enabled projects.

How to use

You connect an MCP client (such as Claude Desktop, Cursor, or Windsurf) to Hi-AI as your MCP server. Use natural language prompts in Korean or English to trigger the appropriate tools for memory management, code analysis, thinking workflows, quality checks, planning, and prompt engineering. When you describe your goal, Hi-AI selects relevant tools, traverses memories in the knowledge graph, analyzes code, and suggests concrete actions.

How to install

Prerequisites you need before installing Hi-AI: Node.js 18.0 or newer, TypeScript 5.0 or newer, an MCP-compatible client, and Python 3.x if you plan to analyze Python code.

Install Hi-AI via npm (global or local) or use the Smithery one-click installer.

# Global installation
npm install -g @su-record/hi-ai

# Local installation
npm install @su-record/hi-ai

MCP server configuration and interop

Hi-AI can be run alongside Hi-GCloud to enable automated error analysis and remediation workflows. Use the following configuration to start both MCP servers and enable automatic integration.

{
  "mcpServers": {
    "hi-ai": {
      "command": "npx",
      "args": ["-y", "@su-record/hi-ai"]
    },
    "hi-gcloud": {
      "command": "npx",
      "args": ["-y", "@polin-go/hi-gcloud"]
    }
  }
}

Operational flow with Hi-GCloud integration

When Hi-GCloud detects issues, it can automatically route problems to Hi-AI for analysis, symbol location, and improvement recommendations, then save the solution in memory to prevent recurrence.

Tools and memory usage reminders

Keep your memory graph updated to reflect changes in your project. Use memory-related tools to save, recall, and relate memories, and use the graph-based search to find related information across time.

Available tools

save_memory

Store information into long-term memory with optional categorization and prioritization.

recall_memory

Retrieve stored memories by key, context, or semantic search to reuse information.

link_memories

Create semantic connections between memories to form a knowledge graph.

get_memory_graph

Query and visualize the knowledge graph to explore relationships.

search_memories_advanced

Perform advanced searches using multiple strategies across memories.

create_memory_timeline

Generate a time-ordered timeline of memory entries for historical context.

prioritize_memory

Set and adjust memory priorities to influence recall and retrieval.

find_symbol

Locate symbol definitions across a codebase.

find_references

Identify all references to a given symbol throughout the project.

analyze_dependency_graph

Analyze code dependencies and visualize relationships including cycles.

analyze_complexity

Compute code complexity metrics such as cyclomatic, cognitive, and Halstead.

validate_code_quality

Evaluate code quality against defined rules and scoring systems.

check_coupling_cohesion

Assess module coupling and cohesion to improve structure.

suggest_improvements

Provide actionable refactoring and quality improvements.

apply_quality_rules

Apply predefined quality rules to codebases.

get_coding_guide

Retrieve coding guidelines and best practices.

generate_prd

Automatically generate Product Requirements Documents from prompts.

create_user_stories

Create user stories with acceptance criteria.

analyze_requirements

Analyze requirements for feasibility and risk.

feature_roadmap

Generate a step-by-step product roadmap.

create_thinking_chain

Produce a thinking chain to structure problem solving.

analyze_problem

Analyze a problem to identify core components and risks.

step_by_step_analysis

Provide a structured step-by-step analysis of a task.

format_as_plan

Transform analyses into an actionable plan.

enhance_prompt

Automatically refine prompts to remove ambiguity.

analyze_prompt

Assess prompt quality for clarity and context.

enhance_prompt_gemini

Optimize Gemini API prompts for better results.

apply_reasoning_framework

Run a 9-step reasoning framework to verify conclusions.

get_usage_analytics

Fetch usage statistics and behavioral analytics.

preview_ui_ascii

Preview a UI in ASCII to help design brainstorming interfaces.

get_current_time

Return the current time for scheduling and logging.