Proposal: Add MCP (Model Context Protocol) Server for AI-Powered GUI Automation #3644

Motivation

Cutter's GUI is powerful but manual. Adding MCP Server support would enable AI Agents to automate analysis workflows:

• Batch processing: AI can analyze multiple binaries without manual GUI interaction
• Report generation: Automated analysis reports with AI-driven insights
• Function annotation: AI can rename functions, add comments, and document code
• Graph generation: Automated call graph and CFG generation for documentation
• Integration: Connect Cutter with AI-powered analysis pipelines

Proposed Implementation

I'd like to contribute an MCP Server module integrated with Cutter's plugin system:

Core Tools (8 tools)

1. open_project - Open/create analysis projects
2. save_project - Save project state
3. get_functions - List all functions with metadata
4. rename_function - Rename functions programmatically
5. add_comment - Add comments to functions/addresses
6. generate_callgraph - Generate call graph (DOT format)
7. generate_cfg - Generate control flow graph
8. export_analysis - Export analysis results

Technical Approach

• Integrate with Cutter's Python API (PythonAPI.h)
• Use CutterPlugin base class for integration
• Leverage rizin backend for analysis
• Implement MCP 1.0 protocol
• Support stdio/SSE/HTTP transports
• Include test suite

Benefits

• Enables AI-assisted reverse engineering workflows
• Automates repetitive analysis tasks
• Generates documentation automatically
• Integrates with existing Cutter plugin system

Request for Feedback

I'd like to:

1. Confirm this fits Cutter's architecture
2. Get feedback on using Python API vs C++ plugin
3. Understand contribution process for plugins
4. Learn about any existing automation efforts

Ready to implement after maintainer approval.

Thanks!