HopperPyMCP

HopperPyMCP - FastMCP Server for Hopper Disassembler

A FastMCP server plugin for the Hopper disassembler that provides powerful analysis tools through the Model Context Protocol (MCP). This plugin allows you to analyze binary files, disassemble procedures, manage documents, and more through AI assistants.

Features

• 🔍 Binary Analysis: Analyze segments, procedures, and data structures
• 🛠️ Disassembly & Decompilation: Get detailed assembly and pseudo-code output
• 📊 Call Graph Generation: Visualize function relationships and program flow
• 🔗 Reference Analysis: Track memory references and cross-references
• 📝 Annotation Tools: Add names, comments, and type information
• 🗂️ Document Management: Handle multiple executable files
• 🔍 String Search: Advanced regex-based string searching

Quick Installation

The installation process automatically detects your Python environment (conda, uv, venv, or system Python) and configures everything for you:

That's it! The script will:

• ✅ Detect your Python environment automatically
• ✅ Install required dependencies (fastmcp)
• ✅ Configure the script with correct Python paths
• ✅ Install to the appropriate Hopper Scripts directory

Supported Environments

• Conda environments (including miniconda/anaconda)
• UV virtual environments
• Python venv/virtualenv
• System Python installations
• macOS and Linux platforms

If you use an environment like conda, uv, or virtualenv, run the install script from within a new environment, since dependencies will be installed by install.py.

Manual Installation Options

Dry Run (Preview Changes)

Force Installation

Uninstallation

Remove the plugin cleanly:

Usage in Hopper

Once installed, the FastMCP server will be available as a script in Hopper.

Starting the Server

After running the script in Hopper, you'll need to launch the MCP server through the Python prompt:

1. First Time Setup - Cache Strings (Recommended)Due to slow Hopper string APIs, the plugin creates optimized string caches for better performance. This process takes about 5-10 minutes per document and saves caches alongside your Hopper document saves.In the Hopper Python prompt, paste:
   cache_strings()
   Wait for caching to complete, then launch the server:
   launch_server()
2. Quick Start (Skip Caching)To start immediately without caching (slower string searches):
   launch_server()
3. Subsequent UsesIf you've already cached strings for your documents:
   launch_server()

The server will run on http://localhost:42069/mcp/ and provide the following tools:

Document Management

• get_all_documents() - Get information about all currently opened documents (Hopper-analyzed binaries)
• get_current_document() - Get information about the current document with its doc_id
• set_current_document(doc_id) - Set the current document by doc_id
• rebase_document(new_base_address_hex) - Rebase the current document to a new base address

Core Analysis Tools

• list_all_segments() - List all segments in the current document with basic information
• get_address_info(address_or_name_list) - Get comprehensive information about multiple addresses/names including segment, section, type, procedure info, and references

Search and Discovery

• search_names_regex(regex_pattern, segment_name, search_type, max_results) - Search for names matching a regex pattern in a specific segment
• search_strings_regex(regex_pattern, segment_name, max_results) - Search for strings matching a regex pattern in a specific segment
• get_string_at_addr(address_hex) - Get the string content at a specific address using the cached strings list

Disassembly & Decompilation

• disassemble_procedure(address_or_name) - Disassemble a procedure into assembly language instructions
• decompile_procedure(address_or_name) - Decompile a procedure to C language code

Call Graph Generation

• get_call_graph(start_addr_hex, direction, max_depth) - Return the call graph starting from a specific address

Name and Symbol Analysis

• get_demangled_name(address_or_name) - Get the demangled name at a specific address or for a given name

Comments and Annotations

• get_comment_at_address(address_hex) - Get the comment at a specific address
• set_comment_at_address(address_hex, comment) - Set a comment at a specific address
• set_name_at_address(address_hex, name) - Set a name/label at a specific address
• mark_data_type_at_address(address_hex, data_type, length) - Mark data type at a specific address

Requirements

• Python 3.8+
• Hopper Disassembler v4 or v5
• FastMCP library (automatically installed)

File Structure

Troubleshooting

Installation Issues

Problem: fastmcp import fails after installation

Problem: Permission denied when writing to Hopper directory

Problem: Wrong Python environment detected

Runtime Issues

Problem: Script not appearing in Hopper

• Verify installation path is correct for your platform
• Check Hopper Scripts directory exists and is readable
• Restart Hopper after installation

Problem: Import errors when running in Hopper

• The installation should handle Python path configuration automatically
• If issues persist, check that the installed script has the correct paths

Platform-Specific Notes

macOS: Scripts install to ~/Library/Application Support/Hopper/Scripts/ Linux: Scripts install to ~/GNUstep/Library/ApplicationSupport/Hopper/Scripts/

Development

Running Tests

Development Installation

For development, you might want to symlink instead of copy:

Support

For issues and questions:

1. Check the troubleshooting section above
2. Review the test files for usage examples
3. Open an issue on the project repository

Note: This plugin requires Hopper's built-in Python interpreter. The installation script automatically configures the necessary Python paths for seamless integration.