MCP Smart Function Pipeline Server

🌐 Watch it now: Video 👆

A Model Context Protocol (MCP) server that uses local LLM (Ollama with Mistral 7B) to interpret natural language queries and execute structured function call pipelines.

🚀 Features

• Natural Language Processing: Uses Ollama with Mistral 7B to parse user queries
• Function Pipeline Execution: Orchestrates complex multi-step workflows
• 50+ Built-in Functions: Invoice management, data processing, email, file operations, etc.
• RESTful API: Easy integration with web interfaces and external systems
• Data Flow Management: Automatic variable resolution and output chaining
• Execution Planning: Dry-run capability to validate pipelines before execution
• Error Handling: Comprehensive error handling and logging

🏗️ Architecture

User Query → AI Model (Mistral 7B) → Function Planner → Execution Engine → Results
                ↓
         Function Library (50+ functions)

📋 Prerequisites

1. Python 3.8+
2. Ollama installed locally with Mistral 7B model

   # Install Ollama (https://ollama.ai/)
   ollama pull mistral:7b

🛠️ Installation

1. Clone the repository

   git clone <repository-url>
   cd codemate-smart-function-pipeline

2. Create virtual environment

   python -m venv env
   # Windows
   env\Scripts\activate
   # Linux/Mac
   source env/bin/activate

3. Install dependencies

   pip install -r requirements.txt

4. Configure environment

   cp .env.example .env
   # Edit .env file as needed

Usage

Starting the Server

# Basic start
python main.py

# With custom configuration
python main.py --host 0.0.0.0 --port 8080 --model mistral:7b

# Debug mode
python main.py --debug

Using the Client

# Run the interactive test client
python client_example.py

GET /functions

3. Execute Query

4. Plan Query (Dry Run)

POST /plan
Content-Type: application/json

{
  "query": "Get invoices for January and calculate total"
}

5. Execute Function Plan

POST /execute-plan
Content-Type: application/json

{
  "function_calls": [
    {"function": "get_invoices", "inputs": {"month": "March"}},
    {"function": "summarize_invoices", "inputs": {"invoices": "$output_0.invoices"}}
  ]
}

🔄 Data Flow Example

User Query: "Get March invoices, summarize them, and email the summary"

Generated Plan:
[
  {"function": "get_invoices", "inputs": {"month": "March"}},
  {"function": "summarize_invoices", "inputs": {"invoices": "$output_0.invoices"}},
  {"function": "send_email", "inputs": {"content": "$output_1.summary", "recipient": "user@example.com", "subject": "Invoice Summary"}}
]

Execution:
1. get_invoices("March") → output_0: {"invoices": [...]}
2. summarize_invoices(output_0.invoices) → output_1: {"summary": {...}}
3. send_email(output_1.summary, "user@example.com", "Invoice Summary") → output_2: {"status": "sent"}

🛡️ Error Handling

• Validation: Function calls are validated before execution
• Graceful Failures: Partial execution results are preserved
• Detailed Logging: Comprehensive execution logs
• Fallback Parsing: Backup parsing if AI model fails

🔧 Configuration

Environment Variables

Variable	Description	Default
OLLAMA_HOST	Ollama server URL	http://localhost:11434
OLLAMA_MODEL	Model name	mistral:7b
SERVER_HOST	Server bind address	localhost
SERVER_PORT	Server port	5000
DEBUG_MODE	Enable debug mode	false

Long Query:

Retrieve all invoices for March, summarize them, and send the summary to shashank.pandey_cs22@gla.ac.in email.