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

Category: security

Tags: prompt-injection, email-security, html-sanitization, local-first, ai-agents, mcp

Score: 6.8/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 6)

InBox Wall is a local-first email sanitizer that strips hidden text, invisible Unicode, and prompt-injection patterns from HTML emails before they reach AI agents. It uses a fail-closed pipeline with dual-database isolation, a REST API, and an MCP server, targeting a real security gap in AI email assistants.

Target audience: AI developers, security engineers, backend developers building email-integrated AI agents

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

View on Hacker News