Show HN: InBox Wall – A local-first, open-source email sanitizer for AI agents

Category: security

Tags: email-security, prompt-injection, local-first, sanitization, ai-agent, privacy

Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)

InBox Wall is a local-first sanitizer that cleans email HTML before feeding it to AI assistants, protecting against hidden text and prompt injection attacks. It combines CSS inlining, inheritance-aware parsing, Unicode normalization, and a fail-closed architecture, then serves safe text via a REST API and MCP server for AI consumption. The project stands out for its thorough threat model, audit dashboard, and privacy-focused design.

Target audience: AI developers, security engineers building AI email integrations, and privacy-conscious users of AI assistants

Repository: https://github.com/Annora998/inboxwall · Python · 1 stars

View on Hacker News