VibeGuard

VibeGuard is an open-source MITM proxy for AI coding privacy. It sits between an IDE or CLI and an AI API, automatically redacting sensitive data in requests and restoring it in responses. The page describes it as built for developers who care about privacy without sacrificing productivity.

Its features include process-level proxying, which only intercepts traffic from targeted processes and leaves the rest of the system network untouched. It is designed to work out of the box with sensible defaults, so installation and running do not require YAML editing. Rules can be changed through an admin UI and take effect immediately without restarting the proxy. The product also stores keywords and placeholder mappings encrypted at rest using the local CA key. Detection is described as multi-layered, combining rule lists, keyword matching, regex, and optional named entity recognition. A browser-based admin dashboard provides real-time audit logs, visual analytics, and rule management.

The workflow shown on the page places VibeGuard between an IDE or CLI and AI services including Claude, GPT, Gemini, Cursor, and Codex. The quick-start instructions mention one-line installation for macOS and Linux, a PowerShell install path for Windows, starting the proxy and trusting the local CA certificate, and launching an AI tool through VibeGuard’s process launcher. Uninstallation commands are also provided for macOS, Linux, and Windows.

The site identifies the project as open source and links an Apache License 2.0 notice, along with GitHub and Docs. It also presents a Chinese and English interface.