Implemented the refactoring proposal to optimize the bridge for better MCP integration:
- Added resources for context loading (instances, functions, disassembly)
- Added prompts for common analysis patterns
- Reorganized tools into namespaced functions for better discoverability
- Implemented current working instance concept for simpler usage
- Added documentation for the namespaces-based approach
This commit fixes an issue where variable operations with paths like
/functions/by-name/FunctionName/variables/varName were not being properly
processed. The handleFunctionResource method now checks for paths that
start with 'variables/' and extracts the variable name for processing.
Added implementation to handle renaming of decompiler-generated variables.
Add documentation for the new strings endpoint in both:
- README.md: List as a feature and add example usage
- GHIDRA_HTTP_API.md: Add complete API reference with parameters and response format
The new endpoint provides:
- GET /strings endpoint with pagination and filtering
- Python bridge support via list_strings() function
- Searching of string data types across memory blocks
- Filtering options for string content
- Add comprehensive description of v2.0.0 features and capabilities
- Update API reference to include all available tools and operations
- Document HATEOAS architecture and response format
- Add detailed examples of using the new data manipulation API
- Update installation instructions for v2.0.0-beta.1
- Update Java plugin version to v2.0.0-beta.1
- Add version identifier to bridge script
- Add API version check in bridge to verify plugin compatibility
- Bridge script will now check for API version 2 compatibility
- Add handleCreateData method to Java plugin to support creating new data
- Add data type mapping to support common types like byte, word, dword, string
- Implement delete_data functionality with graceful handling of missing data
- Add proper error handling when conflicts are detected
- Add comprehensive tests for both create_data and delete_data functionality
- Add comprehensive data manipulation capabilities
- Implement separate pathways for rename-only, type-only, and combined operations
- Fix HTTP request body consumption issue in DataEndpoints
- Standardize on 'type' parameter name instead of 'dataType'
- Add thorough test coverage with dedicated test_data_update.py script
- Update API documentation to version 2 with full endpoint descriptions
- Update CHANGELOG with detailed information about data manipulation features
- Add support for setting data types and renaming data items
- Fix Java API implementation for data operation endpoints
- Create rename_data and set_data_type tools for clearer separation of concerns
- Add comprehensive test scripts for data operations
- Successfully test changing data types and naming
- Add comprehensive test for the set_function_signature tool
- Update test_mcp_client.py with modernized API naming
- Fix HATEOAS link detection to handle both _links and api_links
- Replace getDefinedAddresses with proper Listing methods
- Use getInstructions and getDefinedData instead
- Check both instructions and data for a valid starting address
- Improve fallback strategy for current address retrieval
- Add helper method to handle ReferenceIterator conversion
- Support both Reference[] and ReferenceIterator types
- Replace LocationService with reflection-based approach
- Use listing's defined addresses as fallback
- Fix reference handling using ReferenceIterator
- Use proper Ghidra 11 services for current address retrieval
- Implement location tracking via LocationService and SelectionService
- Java plugin now uses Ghidra ReferenceManager to find real cross-references
- Added detailed information about xrefs, including related functions and instructions
- Bridge script now provides simplified and human-readable text for xrefs
- Support bi-directional search for references to/from addresses
- Added filtering by reference type
- Properly implement getCurrentAddress using Ghidra service APIs
- Remove all legacy API compatibility code
- Implement simplified response structure for AI agents
- Add text representations of structured data like disassembly
- Support addressing functions by name or address
- Improve parameter validation and error handling
- Implement all endpoints from the API specification
- Implemented efficient pagination for variables endpoints to avoid timeout
- Added globalOnly parameter to allow fetching just global variables
- Limited decompilation to only process functions needed for current page
- Improved estimation of total count for better pagination links
- Reduced decompilation timeout to improve performance
- Fixed disassembly endpoint to show real instructions instead of placeholders
- Improved memory endpoint to handle address errors gracefully
- Fixed address arithmetic error in xrefs endpoint
- Added proper error handling and fallbacks in all endpoints
- Add ProgramEndpoints for proper HATEOAS URL structure
- Fix response structure to include required HATEOAS links
- Ensure proper result formats for segments, decompiled functions, and variables
- Reorganize endpoints to use nested resource pattern (/programs/current/functions/{address})
- Fix all tests to ensure HATEOAS compliance
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added comprehensive GHIDRA_HTTP_API.md documenting the Java plugin's endpoints
- Improved bridge_mcp_hydra.py with better docstrings and type hints
- Updated port range calculations to use DEFAULT_GHIDRA_PORT
- Cleaned up comments and simplified code
- Improved error handling and response formatting
- Standardized API response structure across all endpoints
Add toggleCCode, toggleSyntaxTree and setSimplificationStyle controls to
the Java plugin. These allow controlling decompiler output format between
C code (default) and raw decompiler output with syntax trees.
Example usage:
decompile_function_by_address(port=8192, address='0x1000', cCode=True)
decompile_function_by_address(port=8192, address='0x1000', syntaxTree=True)
