Python Kernel Killer

Python Kernel Killer is a VS Code / VSCodium extension (Linux-focused) to inspect and manage running Python-related processes directly from the sidebar.

It is especially useful when Jupyter kernels, multiprocessing jobs, or Python scripts become unresponsive and cannot be stopped from within VS Code.

 

Sidebar with detected processes and tooltip information

Kill process confirmation dialog

---

Features

Process Detection

The extension detects and groups processes into:

• Zombie Python processes
• Orphan Python processes
• IPython / Jupyter kernels
• JupyterLab servers
• Terminal Python processes

Each process shows:

• PID
• CPU usage (%)
• Memory usage (% and RSS in MB)
• Full command

---

Process Control

You can terminate processes directly from the sidebar with three modes:

• Soft kill (SIGTERM)

  kill <PID>

  Graceful shutdown (default, recommended)

• Force kill (SIGKILL)

  kill -9 <PID>

  Immediate termination (use with caution)

---

Safety Mechanisms

To prevent accidental system damage, the extension applies multiple safeguards:

• PID 1 is blocked

  PID 1 (systemd / init) is the root of the process tree on Linux.
  Terminating it can immediately destabilize or shut down the system.

• System Python processes are protected

  Not all Python processes are user workloads. Many desktop and system components are implemented in Python as background services in /usr/bin or /usr/lib.

  • blocked for soft kill
  • require explicit confirmation for force kill

---

Auto Refresh

• Optional auto-refresh every 5 seconds
• Visual indicator when enabled

---

Environment Detection

The extension extracts environment names when possible:

• Conda / Miniforge -> envs/<name>
• venv -> project folder name

Displayed only in the kill confirmation dialog.

---

Security & Privacy

This extension is designed with minimal data exposure.

Instead of collecting all processes, it uses a filtered ps command:

ps -eo pid=,ppid=,stat=,tty=,pcpu=,rss=,comm=,args= | awk '...'

Filtering happens before data reaches the extension, ensuring:

• No full process list is stored
• Only relevant Python/Jupyter processes are processed

---

Requirements

• Linux system
• Standard tools available:
  • ps
  • awk
  • kill

No Python packages required.

---

Usage

1. Open the Python Kernels sidebar
2. Inspect running processes
3. Click a process to terminate it
4. Choose kill mode and confirm action

---

Known Limitations

• Linux only
• Detection based on process patterns (not kernel APIs)
• Remote kernels are only visible if running locally

---

License

MIT

---

Release Notes

See full changelog: 👉 https://github.com/maikfussel/python-kernel-killer/blob/master/CHANGELOG.md

[0.1.0] Latest version

• Process categorization (zombie, orphan, kernels, etc.)
• CPU and memory monitoring
• Kill modes (soft / force)
• Auto-refresh

[0.0.1]

• Initial release