deobf-string

Deobfuscate Strings

Prerequisites

Before starting analysis, check if IDA Pro MCP is connected. Use the MCP tool to verify the connection. If IDA Pro MCP is not available, inform the user:

IDA Pro MCP is not connected. Please open IDA Pro with the MCP plugin loaded and ensure the MCP server is running, then try again.

Workflow

Step 1: Identify Encrypted Strings

Analyze the user-provided code snippet or function to identify string encryption patterns:

• Look for byte arrays or data references that appear to be encrypted strings
• Identify the decryption routine (often called at string usage sites)
• Determine the encryption algorithm by reverse-engineering the decryption function

Common patterns:

• XOR with a single byte or key array
• Custom substitution ciphers
• RC4 or other stream ciphers
• Base64 + XOR combinations
• Stack-constructed strings with arithmetic transformations

Step 2: Analyze the Decryption Routine

Use IDA Pro MCP to:

• Decompile the decryption function
• Identify parameters: encrypted data pointer, length, key
• Determine the algorithm type and complexity

Step 3: Choose Decryption Strategy

Simple algorithms (pure computation, no external dependencies):

• Single-byte or multi-byte XOR
• ADD/SUB/ROT transformations
• Simple lookup tables

→ Write a standalone Python script to decrypt.

Complex algorithms (state-dependent, uses library functions, or too complex to reimplement):

• Custom crypto with many rounds
• Decryption depends on runtime state or global variables
• Algorithm is heavily obfuscated itself

→ Use Unicorn Engine to emulate the decryption function directly from the binary.

Step 4: Implement Decryption

Simple decryption (Python script):

def decrypt_string(enc_bytes, key):
    # Example: XOR decryption
    return bytes([b ^ key[i % len(key)] for i, b in enumerate(enc_bytes)])

Complex decryption (Unicorn emulation):

from unicorn import *
from unicorn.arm64_const import *

def emulate_decrypt(binary_path, func_addr, enc_data, enc_len, key_ptr):
    with open(binary_path, 'rb') as f:
        code = f.read()

    mu = Uc(UC_ARCH_ARM64, UC_MODE_LITTLE_ENDIAN)
    # Map code segment
    mu.mem_map(base_addr, align(len(code)))
    mu.mem_write(base_addr, code)
    # Map stack
    mu.mem_map(stack_base, stack_size)
    mu.reg_write(UC_ARM64_REG_SP, stack_top)
    # Set up arguments (AArch64 calling convention)
    mu.reg_write(UC_ARM64_REG_X0, enc_data_addr)
    mu.reg_write(UC_ARM64_REG_X1, enc_len)
    mu.reg_write(UC_ARM64_REG_X2, key_addr)
    # Emulate
    mu.emu_start(func_addr, func_end)
    # Read decrypted result
    result = mu.mem_read(output_addr, enc_len)
    return bytes(result)

Step 5: Output Results

After decryption, present results to the user:

1. Display decrypted strings with their addresses in a table format:

Address      | Decrypted String
-------------|------------------
0x00412000   | "Hello, World!"
0x00412020   | "/dev/null"

2. Ask the user if they want to generate a patch file:

Decryption complete. Would you like me to write the results to a .patched file that replaces encrypted strings in the binary?

If yes, write a patched binary with encrypted bytes replaced by plaintext (null-terminated), saved as <original_name>.patched.