Awesome Open Source
Awesome Open Source


pyOCD is an open source Python package for programming and debugging Arm Cortex-M microcontrollers using multiple supported types of USB debug probes. It is fully cross-platform, with support for Linux, macOS, and Windows.

A command line tool is provided that covers most use cases, or you can make use of the Python API to enable low-level target control. A common use for the Python API is to run and control CI tests.

Upwards of 70 popular MCUs are supported built-in. In addition, through the use of CMSIS-Packs, nearly every Cortex-M device on the market is supported.

The pyocd command line tool gives you total control over your device with these subcommands:

  • gdbserver: GDB remote server allows you to debug using gdb via either the console or several GUI debugger options.
  • flash: Program files of various formats into flash memory.
  • erase: Erase part or all of an MCU's flash memory.
  • pack: Manage CMSIS Device Family Packs that provide additional target device support.
  • commander: Interactive REPL control and inspection of the MCU.
  • server: Share a debug probe with a TCP/IP server.
  • reset: Hardware or software reset of a device.
  • list: Show connected devices.

The API and tools provide these features:

  • halt, step, resume control
  • read/write memory
  • read/write core registers
  • set/remove hardware and software breakpoints
  • set/remove watchpoints
  • write to flash memory
  • load binary, hex, or ELF files into flash
  • reset control
  • access CoreSight DP and APs
  • SWO and SWV
  • and more!

Configuration and customization is supported through config files, user scripts, and the Python API.


  • PyOCD has moved to its own organization on GitHub! You will continue to see further changes to increase communication and make pyOCD a more independent and open project.

  • Important note: Python 2 support is deprecated and is planned to be dropped from an upcoming release. Existing releases of pyOCD will, of course, continue to work with Python 2. If this is a major problem for you moving forward, please create a new issue describing your concerns. As of 1 January 2021, there has been no negative feedback about dropping Python 2, so this will likely happen sooner rather than later.


  • Python 3.6.0 or later. Currently Python 2.7.9 or later also works, but see note above about Python 2 support being deprecated.
  • macOS, Linux, or Windows 7 or newer
  • A recent version of libusb. See libusb installation for details.
  • Microcontroller with an Arm Cortex-M CPU
  • Supported debug probe
    • CMSIS-DAP v1 (HID), such as:
      • An on-board or standalone debug probe using DAPLink firmware.
      • NXP LPC-LinkII
      • NXP MCU-Link
      • Atmel-ICE
    • CMSIS-DAP v2 (WinUSB), such as:
      • DAPLink firmware version 0254 or newer.
      • Cypress KitProg3 or MiniProg4
      • Keil ULINKplus
    • SEGGER J-Link (experimental)
    • STLinkV2 or STLinkV3, either on-board or the standalone versions.


PyOCD is functionally reliable and fully useable.

The Python API is considered partially unstable as we are restructuring and cleaning it up prior to releasing version 1.0.


The pyOCD documentation is located in the docs directory.

In addition to user guides, you can generate reference documentation using Doxygen with the supplied config file.


The latest stable version of pyOCD may be installed via pip as follows:

$ python3 -mpip install -U pyocd

Note: depending on your system, you may need to use python instead of python3. The same applies if you are using Python 2.7 (but please see the note above about deprecation of Python 2 support in pyocd).

The latest pyOCD package is available on PyPI as well as on GitHub.

To install the latest prerelease version from the HEAD of the master branch, you can do the following:

$ python3 -mpip install --pre -U git+

You can also install directly from the source by cloning the git repository and running:

$ python3 install

Note that, depending on your operating system, you may run into permissions issues running these commands. You have a few options here:

  1. Under Linux, run with sudo -H to install pyOCD and dependencies globally. On macOS, installing with sudo should never be required, although sometimes permissions can become modified such that installing without using sudo fails.
  2. Specify the --user option to install local to your user account.
  3. Run the command in a virtualenv local to a specific project working set.

For notes about installing and using on non-x86 systems such as Raspberry Pi, see the relevant documentation.

libusb installation

pyusb and its backend library libusb are dependencies on all supported operating systems. pyusb is a regular Python package and will be installed along with pyOCD. However, libusb is a binary shared library that does not get installed automatically via pip dependency management.

How to install libusb depends on your OS:

  • macOS: use Homebrew: brew install libusb
  • Linux: should already be installed.
  • Windows: download libusb from and place the .dll file in your Python installation folder next to python.exe. Make sure to use the same 32- or 64-bit architecture as your Python installation. The latest release is available on GitHub; download the .7z archive under Assets. Use the library from the VS2019 folder in the archive.

udev rules on Linux

On Linux, particularly Ubuntu 16.04+, you must configure udev rules to allow pyOCD to access debug probes from user space. Otherwise you will need to run pyOCD as root, using sudo, which is very highly discouraged. (You should never run pyOCD as root on any OS.)

To help with this, example udev rules files are included with pyOCD in the udev folder. The readme in this folder has detailed instructions.

Target support

See the target support documentation for information on how to check if the MCU(s) you are using have built-in support, and how to install support for additional MCUs via CMSIS-Packs.

Standalone GDB server

After you install pyOCD via pip or, you will be able to execute the following in order to start a GDB server powered by pyOCD:

$ pyocd gdbserver

You can get additional help by running pyocd gdbserver --help.

Example command line GDB session showing how to connect to a running pyocd gdbserver and load firmware:

$ arm-none-eabi-gdb application.elf

<gdb> target remote localhost:3333
<gdb> load
<gdb> monitor reset

The pyocd gdbserver subcommand is also usable as a drop in place replacement for OpenOCD in existing setups. The primary difference is the set of gdb monitor commands.

Recommended GDB and IDE setup

The recommended toolchain for embedded Arm Cortex-M development is GNU Arm Embedded, provided by Arm. GDB is included with this toolchain.

For Visual Studio Code, the cortex-debug plugin is available that supports pyOCD.

The GDB server also works well with Eclipse Embedded CDT, previously known as GNU MCU/ARM Eclipse. It fully supports pyOCD with an included pyOCD debugging plugin.

To view peripheral register values either the built-in Eclipse Embedded CDT register view can be used, or the Embedded System Register Viewer plugin can be installed. The latter can be installed from inside Eclipse adding as a software update server URL under the "Help -> Install New Software..." menu item.

Development setup

Please see the Developers' Guide for instructions on how to set up a development environment for pyOCD.


We welcome contributions to pyOCD in any area. Please see the contribution guidelines for detailed requirements for contributions. In order foster a healthy and safe community, we expect contributors to follow the code of conduct.

To report bugs, please create an issue in the GitHub project.


PyOCD is licensed with the permissive Apache 2.0 license. See the LICENSE file for the full text of the license.

Copyright © 2006-2020 Arm Ltd and others (see individual source files)

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
python (49,921
embedded (251
arm (206
debug (139
microcontroller (77
mcu (32
cortex-m (26

Find Open Source By Browsing 7,000 Topics Across 59 Categories