Tekion AI MCP Client

A simple stdio-based Model Context Protocol (MCP) server that provides local Jira and GitLab tools while integrating with a remote MCP server.

Overview

The Tekion MCP Client provides:

• Local Jira Tools: All Jira operations execute locally for better performance and credential security
• Local GitLab Tools: GitLab operations using glab CLI for merge request management
• Remote Server Integration: Forwards other tools to the remote MCP server
• Stdio Interface: Simple stdio transport for direct MCP client connection
• Intelligent Routing: Automatically routes tools between local and remote execution
• Connection Management: Handles remote server connectivity with retry logic

Architecture

[AI Client] ↔ [Tekion MCP Client via stdio] ↔ [Remote MCP Server]
                      ↓
                [Local Jira Tools]
                [Local GitLab Tools]

Features

Local Jira Tools

• get_jira_ticket - Retrieve JIRA ticket information
• create_jira_ticket - Create new JIRA tickets
• create_jira_epic - Create JIRA epics
• add_jira_comment - Add comments to tickets
• link_jira_epic_to_parent - Link epics to parent tickets
• link_jira_story_to_parent - Link stories to parent tickets
• update_jira_ticket - Update existing tickets
• attach_file_to_jira - Attach files to tickets

Local GitLab Tools

• get_gitlab_prs - Get open merge requests for a GitLab project (requires glab CLI)
• save_generated_context - Save generated context files to TekionAI repository via branch and PR (requires glab CLI and git)

Proxy Features

• Automatic tool routing (local vs remote)
• Connection health monitoring
• Retry logic with exponential backoff
• Graceful degradation when remote server unavailable
• Real-time tool discovery from remote server

Prerequisites

GitLab CLI (for GitLab tools)

Install and authenticate the GitLab CLI:

# Install glab CLI
# macOS
brew install glab

# Linux/Windows - see https://gitlab.com/gitlab-org/cli

# Authenticate
glab auth login

Installation

Option 1: Direct Usage with npx (Recommended)

npx @tekionai/mcp-client

Option 2: Local Development

1. Clone and setup:

   cd mcp-client
   npm install

2. Configure environment:

   cp .env.example .env
   # Edit .env with your configuration

3. Build the project:

   npm run build

Configuration

Environment Variables

| Variable | Description | Default | Required | |----------|-------------|---------|----------| | REMOTE_MCP_SERVER_URL | Remote MCP server URL | http://localhost:8085/mcp | No | | TEKIONAI_FRAMEWORK_TYPE | Environment type (DEV/QA) | DEV | No | | TEKIONAI_MODULE | Module name for remote server | - | No | | TEKION_EMAIL | Jira email (local use only) | - | Yes* | | JIRA_TOKEN | Jira API token (local use only) | - | Yes* | | JIRA_BASE_URL | Jira base URL | https://tekion.atlassian.net | No |

*Required for local Jira tools to function. NOT sent to remote server.

Example Configuration

# Remote server
REMOTE_MCP_SERVER_URL=http://your-remote-server:8085/mcp
TEKIONAI_FRAMEWORK_TYPE=DEV
TEKIONAI_MODULE=aec-studio

# Local Jira credentials (NOT sent to remote server)
[email protected]
JIRA_TOKEN=your-jira-api-token

Usage

Starting the Proxy

# Development mode
npm run dev

# Production mode
npm run build
npm start

Connecting AI Clients

Configure your AI client to use the Tekion MCP Client:

Using npx (Recommended)

{
  "mcpServers": {
    "tekion": {
      "command": "npx",
      "args": ["@tekionai/mcp-client"],
      "env": {
        "TEKION_EMAIL": "[email protected]",
        "JIRA_TOKEN": "your-jira-token",
        "REMOTE_MCP_SERVER_URL": "http://your-remote-server:8085/mcp"
      }
    }
  }
}

Using local build

{
  "mcpServers": {
    "tekion": {
      "command": "node",
      "args": ["path/to/mcp-client/dist/index.js"],
      "env": {
        "TEKION_EMAIL": "[email protected]",
        "JIRA_TOKEN": "your-jira-token",
        "REMOTE_MCP_SERVER_URL": "http://your-remote-server:8085/mcp"
      }
    }
  }
}

Testing the Connection

Use any MCP client to test:

# Test with npx
npx @tekionai/mcp-client

# Test local build
node dist/index.js

Tool Routing Logic

The proxy uses intelligent routing to determine where to execute tools:

1. Local Jira Tools: Always executed locally for performance and security
2. Local Tools: Tools registered locally take precedence
3. Remote Tools: Tools not available locally are forwarded to remote server
4. Fallback: If remote server unavailable, returns appropriate error

Error Handling

Remote Server Unavailable

• Local Jira tools continue to work
• Remote tools return connection error
• Automatic reconnection attempts with exponential backoff

Jira Credential Issues

• Clear error messages for missing credentials
• Tools fail gracefully with helpful error information

Network Issues

• Connection retry logic (up to 3 attempts)
• Exponential backoff between retries
• Health check monitoring

Development

Project Structure

src/
├── client.ts              # Main proxy server
├── config.ts              # Configuration management
├── types/                 # TypeScript type definitions
├── proxy/
│   ├── remoteClient.ts    # Remote MCP server client
│   └── toolRouter.ts      # Tool routing logic
└── tools/
    ├── jira/              # Local Jira tools
    │   ├── index.ts
    │   ├── jiraIntegration.ts
    │   └── *.ts           # Individual tool implementations
    └── gitlab/            # Local GitLab tools
        ├── index.ts
        └── *.ts           # Individual tool implementations

Adding New Local Tools

1. Create tool implementation in appropriate directory
2. Export from index file
3. Register in client.ts

Scripts

npm run build      # Build TypeScript
npm run dev        # Development mode with hot reload
npm start          # Start production server
npm run lint       # Run ESLint
npm run format     # Format code with Prettier

Troubleshooting

Common Issues

1. Remote server connection fails

   • Check REMOTE_MCP_SERVER_URL configuration
   • Verify remote server is running and accessible
   • Check network connectivity

2. Jira tools not working

   • Verify TEKION_EMAIL and JIRA_TOKEN are set
   • Test Jira credentials manually
   • Check Jira API token permissions

3. Tools not appearing

   • Check proxy startup logs for registration errors
   • Verify remote server tool discovery
   • Check health endpoint for status

Logs

The proxy provides detailed logging:

• Tool registration
• Routing decisions
• Remote server connectivity
• Error details

Set LOG_LEVEL=debug for verbose logging.

License

MIT License - see LICENSE file for details.