NoviSentinel for VS Code

Scan text for PII, prompt injection, secrets, and toxicity before sending to LLMs — right from your editor.

Requires a running NoviSentinel API (local Docker or the hosted playground).

What it does

@novisentinel chat participant — type @novisentinel <your prompt> in the VS Code chat panel to scan it instantly. See action, risk level, and per-detection explanations inline.
Scan Selection — right-click any selected text → NoviSentinel: Scan Selection, or press Cmd+Shift+S / Ctrl+Shift+S. Detections appear as inline squiggles.
Scan Clipboard — scan whatever's on the clipboard before pasting it into a chat. Press Cmd+Shift+V Cmd+S / Ctrl+Shift+V Ctrl+S.
Status bar indicator — live connection status at a glance. Click to open settings.

Install the extension from the VS Code Marketplace.
Set your API URL in settings: novisentinel.apiUrl (default: http://localhost:8000).
Run NoviSentinel: Set API Key from the Command Palette to store your key securely.
Run docker compose up in your NoviSentinel repo, or point at https://api-play.novisentinel.dev for the public playground.

Command	Keybinding	Description
`NoviSentinel: Scan Selection`	`Cmd+Shift+S` / `Ctrl+Shift+S`	Scan the selected text
`NoviSentinel: Scan Clipboard`	`Cmd+Shift+V Cmd+S`	Scan clipboard contents
`NoviSentinel: Set API Key`	—	Store your API key in the OS keychain
`NoviSentinel: Test Connection`	—	Verify the API is reachable
`NoviSentinel: Open Settings`	—	Open the extension settings

Setting	Default	Description
`novisentinel.apiUrl`	`http://localhost:8000`	Base URL of your NoviSentinel API
`novisentinel.scanContext`	`input`	Treat scanned text as LLM `input` or `output`
`novisentinel.healthPollSeconds`	`30`	Status bar health check interval (seconds, min 5)