Skip to content

usage_examples.md

Latest commit

 

History

History
164 lines (121 loc) · 4.6 KB

usage_examples.md

File metadata and controls

164 lines (121 loc) · 4.6 KB

Mythril MCP Server Usage Examples

This guide shows practical examples of using the Mythril MCP server with different clients.

Example 1: Analyze a Smart Contract

Once you've configured an MCP client (like Claude Desktop or Cline), you can analyze smart contracts using natural language:

Prompt to Client:

Please analyze this Solidity smart contract for security vulnerabilities:

pragma solidity ^0.8.0;

contract SimpleVulnerable {
    address public owner;
    
    constructor() {
        owner = msg.sender;
    }
    
    function destroy() public {
        selfdestruct(payable(msg.sender));
    }
}

What Happens:

  1. The MCP client sends a call_tool request to Mythril MCP server
  2. The server invokes the analyze_contract tool
  3. Mythril performs symbolic execution to find vulnerabilities
  4. Results are returned to the client, which presents them to you

Expected Output:

The analysis will identify the unprotected selfdestruct vulnerability (SWC-106).

Example 2: Disassemble Bytecode

Prompt to Client:

Can you disassemble this EVM bytecode?

0x608060405234801561001057600080fd5b50600080546001600160a01b0319163317905560d1806100316000396000f3fe

What Happens:

The MCP server uses the disassemble_contract tool to convert the bytecode into human-readable assembly instructions.

Example 3: List Available Detectors

Prompt to Client:

What security detectors are available in Mythril?

What Happens:

The MCP server calls list_detectors and returns information about all available security analysis modules.

Example 4: Advanced Analysis with Parameters

Prompt to Client:

Analyze this contract with increased depth for more thorough checking:

[paste contract code]

Use max_depth of 256 and execution_timeout of 7200 seconds.

What Happens:

The MCP client passes additional parameters to the analyze_contract tool, allowing for more thorough analysis.

Example 5: Analyze an On-Chain Contract

Prompt to Client:

Analyze the smart contract at address 0x1234567890123456789012345678901234567890 
on Ethereum mainnet, including its onchain storage data.

What Happens:

The MCP server fetches the contract from the blockchain and analyzes it with enable_onchain_data set to true.

Testing Without a Client

You can test the MCP server manually using a simple Python script:

import json
import subprocess

def test_mythril2_mcp():
    # Start the MCP server
    proc = subprocess.Popen(
        ["mythril2-mcp-server"],
        stdin=subprocess.PIPE,
        stdout=subprocess.PIPE,
        stderr=subprocess.PIPE,
        text=True
    )
    
    # Send initialize request
    init_msg = {
        "jsonrpc": "2.0",
        "id": 1,
        "method": "initialize",
        "params": {
            "protocolVersion": "2024-11-05",
            "capabilities": {},
            "clientInfo": {"name": "test", "version": "1.0"}
        }
    }
    
    proc.stdin.write(json.dumps(init_msg) + "\n")
    proc.stdin.flush()
    
    # Read response
    response = proc.stdout.readline()
    print(json.loads(response))

if __name__ == "__main__":
    test_mythril2_mcp()

Common Use Cases

Security Audit Workflow

  1. Initial Scan: "Analyze this contract for vulnerabilities"
  2. Deep Dive: "Run a more thorough analysis with max_depth=256"
  3. Specific Issues: "Check for reentrancy vulnerabilities specifically"

Development Workflow

  1. Quick Check: Analyze during development to catch issues early
  2. Pre-commit: Use as part of your git pre-commit hooks
  3. CI/CD: Integrate into continuous integration pipelines

Learning and Research

  1. Understand Bytecode: "Disassemble this bytecode to understand what it does"
  2. Compare Patterns: Analyze similar contracts to identify patterns
  3. Detector Exploration: "What detectors would catch this type of vulnerability?"

Tips for Best Results

  1. Provide Context: Include compiler version and any relevant information
  2. Be Specific: If looking for specific vulnerability types, mention them
  3. Iterate: Start with default settings, then adjust timeouts/depth as needed
  4. Review Results: Always review the analysis results carefully

Troubleshooting

Analysis Takes Too Long

Try reducing max_depth or execution_timeout:

Analyze with max_depth=64 and timeout of 600 seconds

Missing Dependencies

If you get errors about missing Solidity compiler:

pip install py-solc-x

Connection Issues

Check that the MCP server is properly configured in your client's settings and that the mythril2-mcp-server command is in your PATH.