Fixing `ctypes` ImportError in Azure Pipelines: Missing `_type_` Attribute

When running pip install in an Azure Pipelines self-hosted agent, you might encounter the following error:

File "~\myselfhostedagent\_work\_tool\Python\3.12.3\x64\Lib\site-packages\pip\_vendor\platformdirs\windows.py", line 254, in _pick_get_win_folder
    import ctypes  # noqa: PLC0415
    ^^^^^^^^^^^^^
  File "ctypes\__init__.py", line 157, in <module>
AttributeError: class must define a '_type_' attribute

Interestingly, if you manually execute pip install using the same python.exe, it works fine. However, when executed within the Azure Pipelines agent, it fails.

In this post, we’ll explore why this happens and how to fix it.

🚨 Why Does This Happen?

1. Python’s DLL Handling in Windows

A typical Python installation on Windows consists of:

python.exe in <PYTHON_ROOT>.
Standard libraries in <PYTHON_ROOT>\Lib.
DLLs (such as _ctypes.pyd, _queue.pyd) in <PYTHON_ROOT>\DLLs

Before Python 3.8, Windows would automatically find these DLLs when needed. However, since Python 3.8, Microsoft tightened security, and Python now explicitly controls how DLLs are loaded using os.add_dll_directory().

2. The Azure Pipelines Environment Is Different

When you run python.exe manually, your user session automatically resolves the DLLs directory.
When the Azure Pipelines agent runs Python, it starts in a minimal environment, and Windows does not find the required DLLs unless they are explicitly referenced.

3. Why pip Fails

The ctypes module relies on system DLLs to function. Since the Python DLLs directory is not in PATH, Windows fails to load necessary components, resulting in:

AttributeError: class must define a '_type_' attribute

✅ How to Fix It

To ensure the agent can find the DLLs, prepend the DLLs directory to the PATH in the pipeline:

steps:
  - script: |
      PYTHON_ROOT=$(python -c "import sys; import os; print(os.path.dirname(sys.executable))")
      echo "##vso[task.prependpath]${PYTHON_ROOT}/DLLs"      
    displayName: "Fix Python DLL Path"

This tells Windows to search Python’s DLLs directory when loading system dependencies.

🔧 Minimal Azure Pipelines YAML File

Here’s a minimal example including UsePythonVersion@0 to ensure the correct Python version is used:

trigger: main

pool:
  name: Self-Hosted-Agent-Pool

steps:
  - task: UsePythonVersion@0
    inputs:
      versionSpec: '3.12'
      addToPath: true

  - script: |
      PYTHON_ROOT=$(python -c "import sys; import os; print(os.path.dirname(sys.executable))")
      echo "##vso[task.prependpath]${PYTHON_ROOT}/DLLs"      
    displayName: "Fix Python DLL Path"

  - script: |
      python -m pip install --upgrade pip
      python -m pip install -r requirements.txt      
    displayName: "Install Dependencies"

📌 Alternative: Use ``

If you prefer not to modify PATH, you can explicitly set the DLL directory inside your Python script before using ctypes:

import os
import sys

dll_path = os.path.join(os.path.dirname(sys.executable), "DLLs")
os.add_dll_directory(dll_path)

import ctypes  # This should now work correctly

However, setting PATH in the pipeline is a more universal fix, ensuring all Python processes in the build agent work without modification.

🎯 Conclusion

Why does this happen? Windows no longer automatically searches Python’s DLLs folder.
Why only in Azure Pipelines? The agent runs with a minimal environment that lacks the required paths.
How to fix it? Prepend the DLLs directory to PATH in the pipeline or use os.add_dll_directory() in Python scripts.

By applying this fix, you can prevent pip install and ctypes-related failures in your Azure Pipelines self-hosted agent.

Happy coding! 🚀

Fixing ctypes ImportError in Azure Pipelines: Missing _type_ Attribute