MaskWall is a data protection gateway developed by Duvaxa. It acts as an intermediary layer between an organization and large language model (LLM) services such as Claude, ChatGPT, and Gemini. Its primary function is to detect and mask sensitive information, including API keys, passwords, personally identifiable information (PII), and internal data, before it is sent to an AI provider. The original values are then restored in the AI's response, allowing the user to interact with the AI without exposing sensitive data to the external service. The tool can be deployed as a self-hosted gateway using a Docker image, configured by setting a single environment variable for API-based usage, or through a Chrome extension for web interfaces. It supports a wide range of AI providers, including OpenAI, Azure OpenAI, Mistral, Groq, and local models. MaskWall identifies credentials, tokens, and PII such as emails, phone numbers, and national IDs from several regions, verified using official checksums. It also allows administrators to define custom patterns for internal project names or client data. The platform includes an admin dashboard for managing API keys, budgets, and usage, and generates a tamper-evident audit log that stores hashed fingerprints rather than raw sensitive values. Alerts can be sent to Slack or Microsoft Teams for specific events, such as blocked requests or budget limits. Additional capabilities include detecting prompt injection, flagging confidential documents, enforcing rate limits per key, and integrating with Claude Code to prevent secrets from appearing in prompts or code. MaskWall is designed for security and compliance teams who need to enforce data governance policies around AI usage. The typical workflow involves installing the gateway, defining a masking policy, and directing applications to route their AI traffic through it. The system can operate in a dry-run monitor mode to observe potential data leaks without blocking them. It provides monthly reports detailing what data was caught, by whom, and the associated spend. The platform is intended for organizations with multiple teams, brands, or client domains, and offers features like per-team policies and exact matching against a customer list using keyed hashes. Duvaxa offers a free walkthrough on the user's own data and paid pilots before commitment, with pricing based on team size and region.