PrivAiTe: Open-Source Proxy Scrubs PII From Claude Code and Codex Traffic
JeremyCMorgan · x · 2026-08-21
PrivAiTe is a self-hosted, open-source LLM proxy that sits on the wire and redacts names, emails and secrets from messages, tool-call arguments, and multimodal content before requests reach the provider, restoring them in replies. It works with Claude Code, Codex, and any OpenAI-compatible app, with fully local detection and zero telemetry.
The author's point: telling a coding agent "never reveal my secrets" is not an egress control. In a test, Claude Code—explicitly instructed to report config variable names but never values—still sent 3 of 4 secrets to its provider; over one session, 23 of 24 planted sensitive values leaked. The project documents its benchmark, threat model, and known misses.
More from coding & agent
- Grok Build to add task scheduling for cloud agents — testingcatalog · 2026-08-21
- Google launches interactive tutorials for Agent architectures with runnable code — fhinkel · 2026-08-21
- Open-source AIUsage: one dashboard for quotas, costs and accounts across 12+ AI subscriptions — tom_doerr · 2026-08-21
- Stanford researcher: automated orchestrators are a big unlock, current versions not there yet — anshulkundaje · 2026-08-21
- CopilotKit open-sources OpenBot: AI coworkers that each get their own computer — Roger_M_Taylor · 2026-08-21
- Ornith 1.5 35B Q5 runs locally inside GitHub Copilot on a Mac M3 Max — DanWahlin · 2026-08-21