DevPulse

Real-time system, container, and Kubernetes pod resource monitoring — GPU, CPU, RAM, VRAM — right in your VS Code sidebar.

If you find DevPulse useful, please consider supporting development via GitHub Sponsors or PayPal!

Why DevPulse?

Working on a shared GPU server? Running dozens of containers or a Kubernetes / k3s cluster? DevPulse gives you a live dashboard without leaving your editor. See who's using what GPU, which container or pod is eating your RAM, attribute GPU processes to the exact pod, and manage them — all from the sidebar.

Features

Container Resources Table

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.

GPU Monitoring

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

System Metrics

CPU and RAM usage displayed in the status bar and sidebar with visual bars. Always visible at a glance while you work.

Container Management

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

Kubernetes / k3s Support

DevPulse monitors Kubernetes pods alongside Docker containers — no separate tool needed. Works with any kubectl-reachable cluster (k3s, k8s, etc.).

• Pods in the Container Resources table — pods appear next to Docker containers with a ☸ badge and namespace, showing CPU/RAM (via kubectl top / metrics-server). Use the Source toggle to filter All / Containers / Pods.
• GPU → pod attribution — GPU processes running inside pods are attributed to the exact namespace/pod, just like containers (matches container cgroups to pod IDs).
• Pod Manager sidebar section — pods grouped by namespace. Expand a pod to see its ports and start a kubectl port-forward to open them in your browser.
• Pod actions — Restart (rollout restart for managed workloads), Stop (scale --replicas=0), Force-delete, Logs, Exec, and Describe — right-click any pod.
• Node-scoped by default — shows only pods running on this machine (so GPU/process attribution works); switch to cluster-wide with dockerMonitor.kubernetes.scope. Auto-disables when kubectl is missing or the cluster is unreachable.
• Tells you when pods are missing — instead of silently showing nothing, the sidebar explains why (kubectl not on PATH, cluster unreachable, no permission to list pods, or everything filtered out by node scope) and links to a full diagnostics report. Docker-only machines never see this — it appears only where kubectl, a kubeconfig, or an in-cluster token exists. Turn it off with dockerMonitor.kubernetes.showWarnings.
• Check Kubernetes on demand — the panel's ... menu (and the Command Palette) has DevPulse: Check Kubernetes, which re-probes from scratch and tells you where you stand: Kubernetes is working — 36 pods visible, or kubectl not found on PATH with what to do about it. It answers even on machines with no Kubernetes at all, which is where the passive warning deliberately stays quiet.

Smart Alerts (Opt-in)

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 notifyOnlyMyContainers to 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

WebView Dashboard

Full GPU/process visualization panel with detailed charts. Open it via Ctrl+Shift+G or Command Palette → "Open GPU Monitor".

Quick Commands

Graceful Degradation

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 Support

No GPU? Shows system metrics and containers only. No Docker? Shows GPU and system metrics. Everything degrades gracefully.

Installation

VS Code Marketplace

Search for DevPulse in the Extensions panel, or:

ext install ANISOFT.devpulse

From VSIX

code --install-extension devpulse-*.vsix

From Source

git clone https://github.com/yusufani/devpulse-monitor.git
cd devpulse-monitor
npm install
npm run package
code --install-extension dist/devpulse-*.vsix

Configuration

Requirements

• 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.

Links

• GitHub Repository
• VS Code Marketplace
• Report Issues
• Sponsor

License

Source Available — free for personal and non-commercial use. Commercial use requires written permission from ANISOFT. See LICENSE for details.