@jvittechs/jai1-mcp

MCP (Model Context Protocol) proxy for connecting IDE AI agents to Jai1 developer tools.

Overview

This package provides a stdio MCP server that proxies requests to the remote Jai1 MCP Server. It allows AI agents in Cursor, Windsurf, and other MCP-compatible IDEs to access powerful developer tools:

• Feature Planning - Get structured development plans with requirements and task breakdowns
• Code Review - Quality analysis with best practices suggestions
• Security Audit - Vulnerability detection and security recommendations
• Commit Messages - Generate Conventional Commits formatted messages
• Tech Stack Suggestions - Get technology recommendations for your project
• Technical Research - Research topics with structured summaries
• Mermaid Fix - Detect and repair Mermaid flowchart syntax errors
• Translation - Professional translation between Vietnamese, Japanese, and English

Installation

npm install -g @jvittechs/jai1-mcp

Or use directly with npx (recommended):

npx @jvittechs/jai1-mcp

Configuration

Required Environment Variables

| Variable | Description | |----------|-------------| | JAI1_ACCESS_KEY | Your Jai1 access key | | JAI1_MCP_SERVER_URL | MCP server URL (provided by your admin) |

Optional Environment Variables

| Variable | Default | Description | |----------|---------|-------------| | JAI1_MCP_TIMEOUT | 120000 | Request timeout in milliseconds | | JAI1_MCP_DEBUG | false | Enable debug logging |

IDE Setup

Cursor IDE

Add to your Cursor MCP configuration (~/.cursor/mcp.json or project .cursor/mcp.json):

{
  "mcpServers": {
    "jai1": {
      "command": "npx",
      "args": ["@jvittechs/jai1-mcp"],
      "env": {
        "JAI1_ACCESS_KEY": "your_access_key_here",
        "JAI1_MCP_SERVER_URL": "https://mcp.your-company.com"
      }
    }
  }
}

Windsurf IDE

Add to your Windsurf MCP configuration:

{
  "mcpServers": {
    "jai1": {
      "command": "npx",
      "args": ["@jvittechs/jai1-mcp"],
      "env": {
        "JAI1_ACCESS_KEY": "your_access_key_here",
        "JAI1_MCP_SERVER_URL": "https://mcp.your-company.com"
      }
    }
  }
}

Claude Desktop

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "jai1": {
      "command": "npx",
      "args": ["@jvittechs/jai1-mcp"],
      "env": {
        "JAI1_ACCESS_KEY": "your_access_key_here",
        "JAI1_MCP_SERVER_URL": "https://mcp.your-company.com"
      }
    }
  }
}

Available Tools

Once connected, the following tools are available to your AI agent:

plan_feature

Create a structured development plan for a feature.

Input:

• feature_description (required): Description of the feature
• codebase_context: Current codebase context
• tech_stack: Tech stack being used
• constraints: Any constraints to consider

suggest_techstack

Get tech stack recommendations based on requirements.

Input:

• requirements (required): Project requirements and goals
• project_type: Type of project (web app, mobile, API, etc.)
• constraints: Budget, team skills, timeline
• preferences: Technology preferences

research

Research a technical topic.

Input:

• topic (required): The topic to research
• context: Context for the research
• depth: One of "brief", "standard", "deep"

review_code

Review code for quality and best practices.

Input:

• code (required): The code or diff to review
• language: Programming language
• scope: Areas to focus on
• project_guidelines: Project-specific guidelines

review_security

Audit code for security vulnerabilities.

Input:

• code (required): The code or diff to audit
• language: Programming language
• context: Context about the code

generate_commit

Generate a commit message from a git diff.

Input:

• diff (required): Git diff content
• files_changed: List of files changed

mermaid_fix

Analyze and fix Mermaid flowchart syntax errors.

Input:

• mermaid_code (required): The Mermaid flowchart code to analyze and fix

Output: Returns JSON with:

• fixed_code: Complete fixed Mermaid code
• explanation: Explanation of fixes
• changes: Array of changes with type, original, fixed, and reason

Example:

{
  "fixed_code": "graph TD\n  A[Start] --> B{Check}\n  B -- \"yes\" --> C[End]",
  "explanation": "Fixed node text with special characters and arrow labels",
  "changes": [...]
}

translate

Professional translation between Vietnamese, Japanese, and English.

Input:

• text (required): Text to translate
• target_language (required): One of "vi", "ja", "en"
• source_language: One of "vi", "ja", "en", "auto" (default: auto)
• context: Additional context (e.g., "UI text", "documentation")

Output: Returns JSON with:

• translated_text: The translated content
• source_language: Detected or provided source language
• target_language: Target language
• notes: Translation notes and special handling

Special Features:

• Handles technical terminology preservation
• Smart number/currency formatting:
  • Vietnamese: 50k → 50.000, man → 10.000
  • Japanese: 万 (man) = 10,000
  • English: 50,000 format
• Context-aware translation for software development

Example:

{
  "translated_text": "My monthly salary is 500,000 Yen",
  "source_language": "vi",
  "target_language": "en",
  "notes": "Converted 'man' from 50 man = 500,000 Yen"
}

Error Codes

| Code | Description | |------|-------------| | ERR-MCP-001 | Invalid access key | | ERR-MCP-002 | Request timeout | | ERR-MCP-003 | Tool not found | | ERR-MCP-004 | Server error | | ERR-MCP-005 | Network error |

Getting an Access Key

To get a JAI1_ACCESS_KEY, please contact the Jai1 team or use the jai1-cli:

# Login to get your access key
jai1 auth login

License

MIT