whereismyai
Inboxwall
Inboxwall is a local-first Python application that sanitizes incoming email HTML before processing by AI.
Inboxwall is a security layer that connects to your email provider via IMAP in read-only mode to fetch recent emails. It passes email HTML through a multi-stage pipeline to remove hidden text, invisible Unicode characters, and known prompt injection patterns. Clean text is saved to a local SQLite database, while blocked threats are quarantined in a separate database. The sanitized text is served to AI tools via a local REST API and a Model Context Protocol server. The tool features dual-database isolation, bearer token authentication, and a dashboard for human review.
Inboxwall at a glance
- Best for: Software engineer
- Runs on: Web, CLI, API, self-hosted, open source
- Pricing: Free
- Released: October 2026
- Category: AI agent tool
- Website: github.com/annora998/inboxwall
Key features
- Sanitizes incoming email HTML for AI assistants
- Detects hidden text using CSS and style analysis
- Strips invisible Unicode characters and homographs
- Isolates blocked threats in a separate database
- Serves clean text via local REST API and MCP server
- Stores IMAP credentials in OS keychain
- Enforces strict bearer token authentication
Use it for
- Protecting AI email assistants from prompt injection
- Sanitizing email HTML before AI summarization
- Quarantining malicious payloads from email inputs
- Auditing blocked email threads and threats
Alternatives to Inboxwall
- AgentMail — Enables AI agents to send and receive emails programmatically via REST API.
- TaintWall — Enforces deterministic security policies at the tool boundary for AI agents.
- Pidgeon — Provides EU-hosted, privacy-focused email inboxes for AI agents with a built-in security firewall.
- SentinelGate — Stops prompt injections and tool hijacks with sub-millisecond overhead.
- HiddenContent.ai — Screens files for concealed instructions before AI assistants act on them.
- Inbin Gate — Prevents unauthorized agent actions by enforcing strict authority checks.
- iFixAi — Audits agent behavior independently without building manual eval harnesses.
- AgentWallet — Enables AI agents to privately pay, message, and trade with zero human intervention.
- Dsh Im — Unifies multiple chat platforms into a single management setting.
- Inillucent — It runs faster than SQLite and includes built-in vector search for AI agents.
- Truth Firewall — Blocks unverified coding agent completion claims automatically.
- Vigil — Drives end-to-end security engagements with specialist agents and a shared Kali container.
All Inboxwall alternatives