agent-email-inbox_skill

This skill helps you securely configure an AI agent inbox with webhooks, tunneling, and prompt-injection protections for real-time email processing.

Python

2.6k

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill agent-email-inbox

SKILL.md41.8 KB

Overview

This skill guides you through configuring a secure email inbox for an AI agent (Moltbot / Clawdbot or similar). It covers inbound email via Resend, webhook endpoint setup, tunneling for local development, and security controls to prevent prompt injection. The goal is reliable, real-time delivery to your agent while minimizing attack surface and operational surprises.

How this skill works

Incoming mail is delivered to Resend (MX) and forwarded to a registered webhook as an event. Your server verifies the webhook signature, fetches full email content, runs security validation rules, and then forwards validated content to the agent (session message or API). Tunneling tools (ngrok, Cloudflare Tunnel) let you expose local endpoints during development; production deployments use serverless or VPS hosting with stable HTTPS.

When to use it

When your agent must react to emails in real time (support tickets, alerts, conversational threads).
When you need a secure, auditable inbound path that avoids polling overhead.
During development to test webhook-driven flows locally before deploying.
When isolating an agent’s email access with domain-scoped API keys is required.
When you want webhook retry guarantees and centralized delivery status.

Best practices

Always verify webhook signatures using the signing secret before processing payloads.
Use domain-scoped API keys or a dedicated Resend account to limit blast radius.
Deploy persistent HTTPS endpoints for production; don’t rely on ephemeral tunnel URLs.
Implement a layered security validation: header checks, sender allowlists, content filters, and prompt-injection detection.
Log webhook deliveries and validation decisions for auditing and incident response.

Example use cases

Support agent that opens a new case and notifies a human and the AI agent on incoming customer emails.
Automated alert handler that escalates critical system emails to the agent immediately.
Conversational email bot that threads replies and maintains session context via webhook-driven messages.
Local development using ngrok or Cloudflare Named Tunnel to validate webhook logic before production deploy.
Sandboxed agent testing with a domain-scoped API key and subdomain mailbox to avoid impacting other projects.

FAQ

Not strictly—free tunnels work for short-term tests, but use a paid static subdomain or Cloudflare named tunnel for stable webhook URLs.

How do I prevent prompt injection via email?

Apply multi-stage validation: verify sender, sanitize and canonicalize content, strip or neutralize executable-like instructions, and run a prompt-injection classifier before sending to the agent.