AI Pulse
Monitor your computer and AI provider usage directly from the Visual Studio Code sidebar.
Features
- System metrics: CPU and RAM usage, plus GPU utilization and VRAM usage when supported.
- AI provider usage: Ollama Cloud, OpenRouter, Claude, and ChatGPT/Codex.
- Automatic refresh: system metrics refresh frequently; provider usage is checked less often to avoid unnecessary network requests.
- Filter and sort: search displayed items and sort them by value or name. Your selection is retained when the view is hidden.
- Theme-aware interface: follows the active VS Code theme, including high-contrast themes.
- Secure credential storage: API keys and tokens are stored in VS Code SecretStorage, not in settings.json.
Provider usage is retrieved from each provider's API. Availability and the fields returned can depend on your account, subscription, API, and token type.
Installation
Install AI Pulse from the Visual Studio Code Marketplace, or install a packaged .vsix file:
- Open the Extensions view (
Ctrl+Shift+X).
- Search for AI Pulse and select Install.
- Select the AI Pulse icon in the Activity Bar to open the panel.
To install a .vsix file, open the Command Palette (Ctrl+Shift+P), run Extensions: Install from VSIX..., and choose the file.
Getting started
- Open the AI Pulse view from the Activity Bar.
- CPU and RAM metrics appear automatically. GPU metrics appear when a supported command-line utility is available.
- To enable provider usage, open the Command Palette and run the matching AI Pulse: Set ... command.
- Select the refresh icon in the view title, or run AI Pulse: Refresh Data, to update immediately.
Open the Command Palette (Ctrl+Shift+P), search for AI Pulse, then select a command:
| Provider |
Command |
Credential |
| Ollama Cloud |
AI Pulse: Set Ollama Cloud API Key |
Create an API key at ollama.com/settings/keys. |
| OpenRouter |
AI Pulse: Set OpenRouter API Key |
Create an API key at openrouter.ai/keys. |
| Claude |
AI Pulse: Set Claude Token |
Use a Claude Code OAuth setup-token (claude setup-token). A claude.ai session key is also supported. |
| ChatGPT/Codex |
AI Pulse: Set ChatGPT Token (Codex) |
By default, AI Pulse reads the Codex CLI access token from %USERPROFILE%\.codex\auth.json. If unavailable, enter an access token with this command. |
The credential input is masked. Enter a new value to replace a credential, or submit an empty value to remove the stored credential. ChatGPT's token source can be changed with aiPulse.chatgpt.tokenSource.
If a provider is not configured, its row explains how to set it up. Provider errors are shown in that provider's row; they do not prevent other metrics from appearing.
GPU metrics
AI Pulse checks for NVIDIA first, then AMD, when GPU detection is set to auto. NVIDIA metrics require nvidia-smi; AMD metrics require amd-smi. If neither utility is installed or available on PATH, GPU and VRAM rows are omitted. Use aiPulse.gpu.detectionOrder to select a vendor or restore automatic detection.
Using the panel
- Filter: type a provider or metric name in the Filter box. Matching is case-insensitive.
- Sort: choose descending value, ascending value, or name A–Z.
- Refresh: use the refresh button in the view title or the
AI Pulse: Refresh Data command.
- Hide rows: use the
aiPulse.display.* settings described below. Hiding a row does not disable collection from other sources.
- Read status colors: green indicates a normal range, yellow a warning range, and red a high-usage range. Thresholds vary by metric. Monetary usage rows display amounts rather than a percentage bar.
Settings
Open Preferences: Open Settings and search for AI Pulse, or edit settings.json directly.
| Setting |
Default |
Description |
aiPulse.refresh.systemSeconds |
3 |
System metric refresh interval, from 1 to 60 seconds. |
aiPulse.refresh.llmMinutes |
10 |
Provider usage refresh interval, from 1 to 1440 minutes. |
aiPulse.gpu.detectionOrder |
auto |
GPU detection: auto (NVIDIA, then AMD), nvidia, or amd. |
aiPulse.display.cpu |
true |
Show or hide the CPU row. |
aiPulse.display.ram |
true |
Show or hide the RAM row. |
aiPulse.display.gpu |
true |
Show or hide the GPU row. |
aiPulse.display.vram |
true |
Show or hide the VRAM row. |
aiPulse.display.ollama |
true |
Show or hide Ollama Cloud rows. |
aiPulse.display.openrouter |
true |
Show or hide OpenRouter rows. |
aiPulse.display.claude |
true |
Show or hide Claude rows. |
aiPulse.display.chatgpt |
true |
Show or hide ChatGPT/Codex rows. |
aiPulse.chatgpt.tokenSource |
codex |
Use the Codex CLI token (codex) or the token stored by the extension (secret). |
Privacy and credentials
- CPU, RAM, and GPU metrics are collected locally.
- When enabled with a credential, provider usage checks send requests to the corresponding provider. Each provider is queried on the configured provider refresh interval and when you request a full refresh.
- Credentials are stored using VS Code SecretStorage. On Windows, VS Code protects secrets using the operating system's credential storage.
- ChatGPT's default
codex token source reads %USERPROFILE%\.codex\auth.json if that file exists. Choose secret to use only the credential stored through the extension.
Troubleshooting
- The view is missing: run
Developer: Reload Window, then select the AI Pulse icon in the Activity Bar.
- A provider asks for a key or token: use the corresponding AI Pulse: Set ... command from the Command Palette.
- A provider shows an HTTP or authentication error: check that the credential is current and valid for that provider, then run AI Pulse: Refresh Data.
- GPU/VRAM rows are missing: confirm that
nvidia-smi or amd-smi is installed and available on PATH, and check aiPulse.gpu.detectionOrder.
- Usage does not update immediately: system metrics and provider usage have separate refresh intervals. Use the refresh command for an immediate full refresh.
- Inspect extension logs: run
Developer: Toggle Developer Tools and check the Console.
Provider APIs and authentication formats may change. If a provider's usage endpoint stops returning data, verify the credential and account access with that provider.
Development
- Open this repository in VS Code.
- Install the project dependencies with
npm install.
- Press
F5 to launch the Extension Development Host.
- In the new window, select the AI Pulse icon from the Activity Bar.
Compile and package the extension:
npm run compile
npx @vscode/vsce package
Install the generated package for local testing:
code --install-extension ai-pulse-0.0.1.vsix
