MCP Agent Tracker

A Model Context Protocol (MCP) server that automatically tracks client-agent conversations without requiring any user interaction.

Features

🗣️ Automatic Conversation Tracking

Zero User Interaction Required: All conversations are tracked automatically
Client Request Logging: Every client prompt/request is logged
Agent Response Logging: Every agent response is captured
Complete Conversation Turns: Full request-response pairs are recorded
Session Management: Automatic session creation and tracking

🔧 MCP Tools Available

get_current_weather(city): Get weather information for a city
agent_interaction(prompt): Interact with the agent
get_interaction_history(limit, session_id): Retrieve conversation history
get_conversation_summary(session_id): Get conversation statistics and patterns

📊 Automatic Monitoring

Background Health Checks: Continuous system monitoring every 5 minutes
Automatic Metadata Collection: System info, process details, uptime
Error Tracking: Comprehensive error logging and recovery
Performance Metrics: Execution times and system health

How It Works

1. Automatic Session Creation

# Sessions are created automatically when the server starts
# No user input required
logger.get_or_create_session()

2. Client Request Tracking

# Every client request is automatically logged
logger.log_client_request(f"Get weather for {city}")

3. Agent Response Tracking

# Every agent response is automatically captured
logger.log_agent_response(response)

4. Complete Conversation Logging

# Full conversation turns are recorded
logger.log_conversation_turn(
    client_request=f"Get weather for {city}",
    agent_response=response
)

5. Background Monitoring

# System health is monitored continuously
# No user interaction needed
def background_monitoring():
    while True:
        logger.log_interaction(interaction_type='health_check', ...)
        time.sleep(Config.MONITORING_INTERVAL_SECONDS)

Configuration

Environment Variables

# Enable/disable features
ENABLE_BACKGROUND_MONITORING=true
MONITORING_INTERVAL_SECONDS=300
ENABLE_AUTOMATIC_METADATA=true

# Database and logging
DATABASE_URL=
DB_PATH=./data/agent_tracker.db
LOG_LEVEL=INFO

Configuration Options

ENABLE_BACKGROUND_MONITORING: Enable continuous system monitoring
MONITORING_INTERVAL_SECONDS: How often to run health checks (default: 300s)
ENABLE_AUTOMATIC_METADATA: Collect system info automatically

Database Schema

AgentInteraction Table

CREATE TABLE agent_interactions (
    id INTEGER PRIMARY KEY,
    timestamp TIMESTAMP,
    session_id VARCHAR(255),
    user_id VARCHAR(255),
    interaction_type VARCHAR(100),  -- 'client_request', 'agent_response', 'conversation_turn'
    prompt TEXT,                    -- Client request
    response TEXT,                  -- Agent response
    status VARCHAR(50),
    error_message TEXT,
    meta_data JSON                  -- Automatic system metadata
);

Session Table

CREATE TABLE sessions (
    id VARCHAR(255) PRIMARY KEY,
    user_id VARCHAR(255),
    started_at TIMESTAMP,
    last_activity TIMESTAMP,
    total_interactions INTEGER,
    meta_data JSON
);

Usage Examples

Basic Conversation Tracking

@mcp.tool()
def my_tool(prompt: str) -> str:
    # Client request is automatically logged
    logger.log_client_request(prompt)

    # Process the request
    response = process_request(prompt)

    # Agent response is automatically logged
    logger.log_agent_response(response)

    # Complete conversation turn is recorded
    logger.log_conversation_turn(prompt, response)

    return response

Getting Conversation History

# Get recent conversations
history = get_interaction_history(limit=10)

# Get conversation summary
summary = get_conversation_summary()

Security Features

Environment Variables: All configuration via environment variables
No Hardcoded Secrets: Secure credential management
Isolated Database Schema: Separate schema for tracking data
Error Isolation: Logging failures don't break main functionality

Getting Started

Copy environment file:
cp env.example .env
Configure your environment:
# Edit .env with your settings ENABLE_BACKGROUND_MONITORING=true MONITORING_INTERVAL_SECONDS=300
Run the server:
python main.py
Monitor conversations:
# Use the MCP tools to interact and track conversations

🚀 Using in Cursor

Prerequisites

Cursor IDE installed on your system
Python 3.8+ with pip/uv package management
Git for cloning the repository

Step 1: Setup MCP Server

Clone and navigate to your project:
cd /path/to/your/mcp/project
Install dependencies:
# Using pip pip install -r requirements.txt # Or using uv (recommended) uv sync
Configure environment:
cp env.example .env # Edit .env with your preferred settings

Step 2: Configure Cursor for MCP

Open Cursor Settings:
- Press Cmd+, (Mac) or Ctrl+, (Windows/Linux)
- Or go to Cursor → Preferences → Settings
Add MCP Configuration:
{ "mcpServers": { "mcp-project": { "command": "python", "args": ["/absolute/path/to/your/project/main.py"], "env": { "PYTHONPATH": "/absolute/path/to/your/project" } } } }
Alternative: Use relative paths (if Cursor is opened in project directory):
{ "mcpServers": { "mcp-project": { "command": "python", "args": ["./main.py"] } } }

Step 3: Test MCP Integration

Restart Cursor after adding MCP configuration
Open Command Palette (Cmd+Shift+P or Ctrl+Shift+P)
Type "MCP" to see available MCP commands
Test a tool:
- Use get_current_weather("New York") to test weather functionality
- Use agent_interaction("Hello, how are you?") to test conversation tracking
- Use get_system_status() to check system health

Step 4: Use MCP Tools in Cursor

Available Tools

get_current_weather(city): Get weather for any city
agent_interaction(prompt): Interact with the agent and track conversations
get_interaction_history(limit, session_id): View conversation history
get_conversation_summary(session_id): Get conversation analytics
get_system_status(): Check system health and configuration
test_conversation_tracking(message): Test the tracking system

Example Usage in Cursor

Open Command Palette (Cmd+Shift+P)
Type MCP command:
MCP: mcp-project: get_current_weather
Enter parameters when prompted:
city: San Francisco
View results in the output panel

Step 5: Monitor and Debug

View Conversation History

# In Cursor terminal or via MCP tools
python -c "
from main import get_interaction_history
print(get_interaction_history(limit=5))
"

Check System Status

# Via MCP tools in Cursor
get_system_status()

Test Conversation Tracking

# Via MCP tools in Cursor
test_conversation_tracking("Test message from Cursor")

Troubleshooting

Common Issues

"MCP server not found":
- Check the absolute path in your Cursor settings
- Ensure the Python path is correct
- Verify the server is running
"Import errors":
- Check PYTHONPATH in MCP configuration
- Ensure all dependencies are installed
- Verify you're in the correct directory
"Permission denied":
- Make sure main.py is executable
- Check file permissions
- Try running with python3 instead of python

Debug Commands

# Test MCP server directly
python main.py

# Check dependencies
pip list | grep mcp

# Verify configuration
python -c "from config import Config; print(Config.ENVIRONMENT)"

Advanced Configuration

Custom MCP Server Names

{
  "mcpServers": {
    "my-custom-mcp": {
      "command": "python",
      "args": ["./main.py"],
      "env": {
        "ENVIRONMENT": "development",
        "LOG_LEVEL": "DEBUG"
      }
    }
  }
}

Multiple MCP Servers

{
  "mcpServers": {
    "mcp-project": { "command": "python", "args": ["./main.py"] },
    "another-mcp": { "command": "python", "args": ["./other_mcp.py"] }
  }
}

Benefits in Cursor

✅ Seamless Integration: Use MCP tools directly in your IDE
✅ Real-time Monitoring: Track conversations as you work
✅ Debugging Tools: Built-in testing and monitoring functions
✅ Performance Insights: Monitor system health and usage
✅ Conversation Analytics: Analyze interaction patterns
✅ Zero Configuration: Automatic setup and tracking

Your MCP server will now be fully integrated with Cursor, providing powerful conversation tracking and monitoring capabilities right in your development environment!

What Gets Tracked Automatically

✅ Client Requests: Every prompt, question, or request
✅ Agent Responses: Every response, answer, or action
✅ Conversation Flow: Complete request-response pairs
✅ System Health: Background monitoring and metrics
✅ Error Handling: All errors and exceptions
✅ Session Data: User sessions and activity
✅ Metadata: System info, timestamps, environment

❌ Tool Usage: Internal MCP tool executions are not tracked
❌ User Input: No manual logging required
❌ Configuration: Automatic setup and management

The system is designed to be completely hands-off - once started, it will track all client-agent conversations automatically without any intervention needed.

This server cannot be installed

security - not tested

license - not found

quality - not tested

How are these scores calculated?

hybrid server

The server is able to function both locally and remotely, depending on the configuration or use case.

Automatically tracks and logs all client-agent conversations in real-time without user intervention. Provides conversation history, analytics, weather tools, and continuous system health monitoring with complete request-response pair recording.

Related MCP Servers

A2A MCP Server
khulnasoft-com
-
security
F
license
-
quality
A centralized server that tracks and manages connected agents, providing a web interface to monitor their status while enabling agent communication through a central point.
Last updated -
MCP Work History Server
nocoo
A
security
A
license
A
quality
A Model Context Protocol server that enables AI tools to automatically log their activities with detailed metrics like timestamps, token usage, and costs into daily markdown worklog files.
Last updated -
1
1
MIT License
prometheus-mcp
idanfishman
A
security
A
license
A
quality
Provides seamless integration between AI assistants and Prometheus, enabling natural language interactions with your monitoring infrastructure. This server allows for effortless querying, discovery, and analysis of metrics.
Last updated -
10
29
20
MIT License
AI Conversation Logger
fablefang
A
security
F
license
A
quality
Enables AI assistants to automatically log and manage conversation history with developers in structured markdown format. Provides powerful search and context suggestions to help AI understand project history and maintain continuity across sessions.
Last updated -
4
1

View all related MCP servers