Whiteout AI - IDE Governance

Enterprise AI governance and compliance for VS Code, Cursor, Windsurf, and Google Antigravity IDE.

Whiteout AI enables organizations to safely adopt generative AI tools while enforcing compliance, data security, and auditability across developer workflows.

Features

Real-Time Prompt Auditing

• Claude Code — monitors ~/.claude/history.jsonl for all prompts
• OpenAI Codex — monitors Codex session files for prompt/response pairs
• Cursor AI — monitors Cursor's built-in AI via workspace state database
• GitHub Copilot — intercept prompts via @whiteout chat participant
• VS Code built-in chat — audits native agent-mode sessions from VS Code's chat session storage
• Google Antigravity — audits the Antigravity app, the Antigravity IDE agent panel and the agy CLI from their agent transcripts
• Gemini CLI — monitors Gemini CLI chat transcripts
• Aider — monitors .aider.chat.history.md in your workspace folders

Policy Compliance

• Scans every AI prompt against your organization's compliance policies
• Instant status bar feedback: Scanning → Approved / Blocked
• Violation notifications with policy details
• Offline queue with automatic retry when backend is unavailable

Prompt Injection Defense

• Real-time detection of prompt injection attacks in AI tool outputs
• Local pattern scanning plus backend enrichment for high-severity threats
• Claude Code, Cursor, OpenAI Codex, and Google Antigravity hook management for tool-level defense
• Codex hooks are registered automatically in ~/.codex/hooks.json. Codex runs them once you trust Whiteout's hooks when it asks at startup; until then the status reads Codex: needs approval. The command is the same on every machine, so you only approve it once
• Whiteout: Turn On Codex Protection registers the hooks machine-wide instead (/etc/codex/requirements.toml, or %ProgramData%\OpenAI\Codex\requirements.toml on Windows), after your OS asks for administrator approval. Codex then runs them for every user with no trust step, and the hooks.json copy is removed. On Linux the command is typed into a terminal for you to review and run with sudo
• Antigravity hooks are registered once in ~/.gemini/config/hooks.json, which the Antigravity app, Antigravity IDE and the agy CLI all share: a prompt gate that runs before every model call and a tool gate that runs before every shell, file, browser and MCP action. When Whiteout Desktop Guard is installed it owns this registration and the extension stands by

Enterprise Features

• Centralized policy management via admin dashboard
• Per-group policy enforcement
• Device heartbeat and session tracking
• Proactive token refresh with sleep/wake detection
• Secure credential storage

Getting Started

1. Install the extension from the VS Code Marketplace (Cursor, Windsurf, and Antigravity IDE install it from Open VSX). On managed devices Whiteout Desktop Guard installs and updates it for you
2. Open the Command Palette (Cmd+Shift+P) and run Whiteout: Sign In
3. Enter your organization email and password
4. Select your policy group via Whiteout: Select Policy Group

The extension begins monitoring immediately — look for the shield icon in the status bar.

Configuration

All settings are available under Settings > Extensions > Whiteout AI Governance:

Supported IDEs

• VS Code — full feature support
• Cursor — built-in AI monitoring + hook-based interception
• Windsurf — Claude Code and terminal monitoring
• Antigravity IDE — Antigravity agent auditing and hook-based defense, plus Claude Code, Codex, and terminal monitoring

Requirements

• An active Whiteout AI organization account
• Network access to your organization's Whiteout AI API endpoint

About

Built by Groovy Security. Whiteout AI is an enterprise platform for safe, compliant AI adoption.