Real-time system and container resource monitoring — GPU, CPU, RAM, VRAM — right in your VS Code sidebar.
If you find DevPulse useful, please consider sponsoring the project to support development!
Working on a shared GPU server? Running dozens of containers? DevPulse gives you a live dashboard without leaving your editor. See who's using what GPU, which container is eating your RAM, and kill runaway processes — all from the sidebar.
Sortable table of all running containers showing VRAM, CPU, RAM, Net I/O, and Disk I/O. Click column headers to sort by any metric. Use the Group by Owner button to see per-user resource totals — great for shared servers where you need to know who's using what.
Full support for NVIDIA (nvidia-smi), AMD (rocm-smi), and Apple Silicon (system_profiler). Each GPU shows:
- VRAM usage with a color-coded visual map per user
- Temperature, power draw, and utilization percentage
- Per-user breakdown: expand any GPU to see which user/container is consuming VRAM
- Multi-GPU summary line when multiple GPUs are present
CPU and RAM usage displayed in the status bar and sidebar with visual bars. Always visible at a glance while you work.
Right-click any container in the sidebar for a full set of actions:
- Stop / Restart / Force Kill — manage container lifecycle
- Exec — open a shell inside the container
- Attach — attach to the container's main process
- Logs — tail the last 100 lines in a VS Code terminal
- Open Port — open exposed ports directly in your browser
- Show Environment Variables — view all env vars set in the container
- Show Volume Mounts — inspect bind mounts and volumes
- Copy Container ID / Name / Image — quick copy to clipboard
All notifications are disabled by default. Enable them with the Alerts toggle button in the container table toolbar. Once enabled, DevPulse watches for:
- VRAM Critical — warns when any GPU exceeds 90% VRAM usage (resets at 85%)
- Container Stopped — notifies when containers disappear between refresh cycles. Use
notifyOnlyMyContainersto filter by your OS user. - Idle GPU — detects GPUs with VRAM allocated but near-zero utilization for 3+ consecutive samples — helps find forgotten jobs
- Memory Leak — flags GPUs with monotonically increasing VRAM over 10 samples with 5%+ growth — early warning for leaky training runs
Full GPU/process visualization panel with detailed charts. Open it via Ctrl+Shift+G or Command Palette → "Open GPU Monitor".
| Shortcut | Command |
|---|---|
Ctrl+Shift+G |
Open GPU Monitor panel |
Ctrl+Shift+D |
Find Container (quick pick) |
Ctrl+Shift+T |
Show Top GPU Consumers |
No GPU? Shows system metrics and containers only. No Docker? Shows GPU and system metrics. Running inside a container? Automatically falls back to docker top for process detection. Everything works with whatever is available.
| Platform | CPU/RAM | GPU | Containers |
|---|---|---|---|
| Linux | /proc | nvidia-smi, rocm-smi | docker CLI |
| macOS | sysctl, vm_stat | system_profiler | docker CLI |
| WSL2 | /proc | nvidia-smi (passthrough) | docker CLI |
No GPU? Shows system metrics and containers only. No Docker? Shows GPU and system metrics. Everything degrades gracefully.
Search for DevPulse in the Extensions panel, or:
ext install ANISOFT.devpulse
code --install-extension devpulse-*.vsixgit clone https://github.com/yusufani/devpulse-monitor.git
cd devpulse-monitor
npm install
npm run package
code --install-extension dist/devpulse-*.vsix| Setting | Default | Description |
|---|---|---|
dockerMonitor.refreshInterval |
10 |
Sidebar refresh interval (seconds) |
dockerMonitor.webviewRefreshInterval |
5 |
WebView refresh interval (seconds) |
dockerMonitor.gpuMonitoring |
true |
Enable GPU monitoring |
dockerMonitor.dockerBinary |
"" |
Custom docker binary path (auto-detected if empty) |
dockerMonitor.enableNotifications |
false |
Enable automatic notifications (VRAM alerts, container stop, idle GPU, memory leak) |
dockerMonitor.notifyOnlyMyContainers |
true |
Only show container stop notifications for containers owned by the current OS user |
- Docker CLI — for container monitoring (optional)
- nvidia-smi — for NVIDIA GPU monitoring (optional)
- rocm-smi — for AMD GPU monitoring (optional)
None of these are strictly required. DevPulse shows whatever is available.
Source Available — free for personal and non-commercial use. Commercial use requires written permission from ANISOFT. See LICENSE for details.