DevOps AI Assistant with Model Context Protocol

A natural language interface for DevOps automation using LLMs and the Model Context Protocol (MCP) to orchestrate Jira, GitHub, and Slack operations.

What is Model Context Protocol?

Model Context Protocol (MCP) is a standardized interface that enables AI models to interact with external tools and services in a consistent way. Key advantages:

• Abstraction: Tools expose their functionality through a uniform protocol, making them easily consumable by AI models
• Composability: Chain multiple tool invocations into complex workflows
• Safety: Built-in validation, rate limiting, and access control
• Extensibility: Easy to add new tools while maintaining the same interface
• Observability: Standard logging and monitoring across all tool interactions

How It Works

This project demonstrates a complete AI assistant for DevOps automation:

1. Natural Language Understanding

   • User submits plain English request
   • GPT converts it into a structured plan of tool invocations
   • Each step includes tool name, action, arguments, and explanatory comment

2. MCP Execution

   • Plan is executed step by step through MCP servers
   • Each server handles specific tool (Jira/GitHub/Slack)
   • Results are collected with success/failure status

3. Result Summarization

   • GPT analyzes execution results
   • Generates human-friendly summary
   • Highlights key outcomes and any errors

Setup Instructions

1. Clone and Configure

   # Clone repository
   git clone https://github.com/your-org/mcp-demo
   cd mcp-demo
   
   # Copy and edit environment variables
   cp backend/app/.env.example backend/app/.env

2. Configure Environment Edit backend/app/.env:

   # OpenAI API key for GPT model access
   OPENAI_API_KEY=your_openai_api_key_here
   
   # Jira MCP Server
   JIRA_MCP_ENDPOINT=https://mcp.example.com/jira
   JIRA_MCP_TOKEN=your_jira_mcp_token_here
   
   # GitHub MCP Server
   GITHUB_MCP_ENDPOINT=https://mcp.example.com/github
   GITHUB_MCP_TOKEN=your_github_mcp_token_here
   
   # Slack MCP Server
   SLACK_MCP_ENDPOINT=https://mcp.example.com/slack
   SLACK_MCP_TOKEN=your_slack_mcp_token_here

3. Run with Docker

   docker-compose up --build

4. Access the UI

   • Frontend: http://localhost:5173
   • Backend API: http://localhost:8000

Example Prompts

1. Cross-Tool Status Report

   Prompt: "Summarize high-priority Jira issues and merged GitHub PRs; post to #devops."
   
   Result:
   - Fetches high-priority Jira issues
   - Lists recently merged PRs
   - Posts formatted summary to Slack
   

2. GitHub Activity Report

   Prompt: "List merged PRs since last week."
   
   Result:
   - Retrieves PRs merged in last 7 days
   - Formats as markdown report
   

3. Automated Task Creation

   Prompt: "Create Jira tickets for PRs with TODO comments."
   
   Result:
   - Scans merged code for TODOs
   - Creates Jira tasks
   - Notifies team on Slack
   

See demo_prompts.json for more examples with detailed plans.

Demo Talking Points

1. Natural Language Power

   • Show how complex workflows are expressed in plain English
   • Highlight plan generation and smart parameter selection
   • Demonstrate error handling and recovery

2. Tool Integration

   • Explain MCP server architecture
   • Show how tools are abstracted and composed
   • Discuss security and access control

3. Developer Experience

   • Demo the approval workflow for reviewing plans
   • Show real-time updates and error reporting
   • Highlight the time saved vs manual operations

4. Architecture Benefits

   • Explain how MCP enables tool independence
   • Discuss scaling and monitoring capabilities
   • Show how to add new tools and actions

5. Real-World Impact

   • Share metrics on time saved
   • Discuss reduction in human error
   • Show examples of complex workflows automated

Contributing

See CONTRIBUTING.md for development setup and guidelines.

License

This project is licensed under the MIT License - see LICENSE for details.