Manticore

Manticore is a prototyping tool for dynamic binary analysis, with support for symbolic execution, taint analysis, and binary instrumentation.

Features

• Input Generation: Manticore automatically generates inputs that trigger unique code paths
• Crash Discovery: Manticore discovers inputs that crash programs via memory safety violations
• Execution Tracing: Manticore records an instruction-level trace of execution for each generated input
• Programmatic Interface: Manticore exposes programmatic access to its analysis engine via a Python API

Scope

Manticore supports binaries of the following formats, operating systems, and architectures. It has been primarily used on binaries compiled from C and C++.

• OS/Formats: Linux ELF, Windows Minidump
• Architectures: x86, x86_64, ARMv7 (partial)

Requirements

Manticore is officially supported on Linux and uses Python 2.7.

Installation

These install instructions require pip 7.1.0, due to --no-binary. If you have an older pip version, you might be able to use --no-use-wheel instead.

We recommend the use of Manticore in a virtual environment, though this is optional. To manage this, we recommend installing virtualenvwrapper. Then, to set up a virtual environment, in the root of the Manticore repository, run

mkvirtualenv manticore

Then, from the root of the Manticore repository, run:

pip install --no-binary capstone .

or, if you didn't use a virtualenv and would like to do a user install:

pip install --user --no-binary capstone .

This installs the Manticore CLI tool manticore and the Python API.

Then, install the Z3 Theorem Prover. Download the latest release for your platform and place the z3 binary in your $PATH.

Note: The --no-binary flag is a workaround for a known Capstone issue that may occur.

For developers

For a dev install, run:

pip install -e --no-binary capstone --no-binary keystone-engine .[dev]

This installs a few other dependencies used for tests which you can run with some of the commands below:

cd /path/to/manticore/
# all tests
nosetests
# just one file
nosetests tests/test_armv7cpu.py
# just one test class
nosetests tests/test_armv7cpu.py:Armv7CpuInstructions
# just one test
nosetests tests/test_armv7cpu.py:Armv7CpuInstructions.test_mov_imm_min

Quick start

Install and try Manticore in a few shell commands:

# follow install instructions in README.md before beginning
cd /path/to/manticore/
cd examples/linux
make
manticore basic
cat mcore_*/*1.stdin | ./basic
cat mcore_*/*2.stdin | ./basic
cd ../script
python count_instructions.py ../linux/helloworld # ok if the insn count is different

Here's an asciinema of what it should look like: https://asciinema.org/a/567nko3eh2yzit099s0nq4e8z

Usage

$ manticore ./path/to/binary  # runs, and creates a directory with analysis results

or

# example Manticore script
from manticore import Manticore

hook_pc = 0x400ca0

m = Manticore('./path/to/binary')

@m.hook(hook_pc)
def hook(state):
  cpu = state.cpu
  print 'eax', cpu.EAX
  print cpu.read_int(cpu.SP)

  m.terminate()  # tell Manticore to stop

m.run()

FAQ

How does Manticore compare to angr?

Manticore is simpler. It has a smaller codebase, fewer dependencies and features, and an easier learning curve. If you come from a reverse engineering or exploitation background, you may find Manticore intuitive due to its lack of intermediate representation and overall emphasis on staying close to machine abstractions.