Rogue_Binary // Triage_Mode
Document_Viewer: /wiki/rbinghidra-setup Back_To_Index

Rogue Binary MCP

Ghidra MCP Setup

Compile and configure the rbinghidra Model Context Protocol server natively.

Section: Requirements

The rbinghidra MCP server requires a headless Ghidra installation and a working Rust toolchain. It runs completely locally without containers.

  • Ghidra installation directory (version 12.1 or newer recommended).
  • Java Development Kit (JDK 21 or newer) configured and available in your environment path.
  • Rust toolchain (Cargo/rustc 1.75 or newer).
Section: Compilation

Clone the repository and compile the native Rust binary inside your local environment.

  • Navigate into the cloned rbinghidra directory.
  • Run the cargo build command with release optimizations.
  • The compiled native executable will be located in the target directory.
Build Native Binary 
                        cd ~/rbinghidra
cargo build --release
Section: MCP Client Configuration

Add the server to your AI agent client (such as Claude Desktop or Claude Code) using absolute executable paths and standard environment variables.

  • Define GHIDRA_INSTALL_DIR to point to your headless Ghidra folder.
  • Define RBM_CACHE_DIR to control where Ghidra project files and decompilation cache are saved.
  • Configure execution to run directly through the native binary pipe.
Claude Desktop Config 
                        {
  "mcpServers": {
    "rbinghidra": {
      "command": "/path/to/rbinghidra/target/release/rbinghidra",
      "env": {
        "GHIDRA_INSTALL_DIR": "/path/to/ghidra",
        "RBM_CACHE_DIR": "~/.cache/rbinghidra"
      }
    }
  }
}
Claude Code Config 
                        claude mcp add rbinghidra -- \
  /path/to/rbinghidra/target/release/rbinghidra \
  -e GHIDRA_INSTALL_DIR=/path/to/ghidra \
  -e RBM_CACHE_DIR=~/.cache/rbinghidra
Section: Cache and Projects

The server manages private Ghidra headless projects automatically inside your specified cache directory. Consecutive calls reuse these pre-imported databases for sub-second responses.

  • Ghidra project databases are persisted under the ghidra/ subpath of your cache directory.
  • In-flight locks ensure multiple queries to the same binary do not cause conflicts.
  • Timeout configurations can be controlled via the RBM_GHIDRA_TIMEOUT environment variable.