Reverse Pi

A customized Pi Agent tailored to my personal reverse engineering workflow.

Quick Start

1. Install Pi on Linux (Linux is required).
2. Review config.toml and adjust the tool paths and ports for your machine.
3. Make the ./pi script executable: chmod +x ./pi.
4. Run ./pi.

Configuration

User-editable tool paths and service settings for the bundled skills live in the repository-level config.toml:

Table	Used by	Keys
[jadx]	manage-jadx-mcp	port, jadx_port, mcp_server, gui, python, gui_start_timeout, state_dir, gui_env_keep
[ghidra]	manage-ghidra-mcp	home, mcp_dir, python, host, port, mcp_host, mcp_port
[idalib]	manage-idalib-mcp	port

Scripts read these values only from config.toml; environment variables do not override them. Values absent from the file fall back to built-in defaults. Path values may start with ~/.

If config.toml is missing, the skill scripts exit with an error. Each skill's scripts/config.sh contains the same inline TOML reader, which supports tables, bare keys, strings, integers, booleans and comments. If you change a service port, update .pi/mcp.json to match.

Features

• Automatically launches reverse engineering tools and supports MCP integration (in progress).
• Standardizes the reverse engineering workflow (in progress).
• Stores session history in the current directory under .pi/sessions/. (easy cleanup of reverse engineering records?)

Planned Features

• Automatic installation of required dependencies. (Maybe it's not necessary?)manage