email-security
por ivaavimusic
Coding Agents & IDEsResumo
Protect AI agents from email-based attacks including prompt injection, sender spoofing, malicious attachments.
Descrição
O que esta skill faz
A security layer for AI agents that process emails. Performs sender verification, content sanitization, and threat detection before any email-based command is executed. Covers prompt injection, sender spoofing, malicious attachments, and social engineering.
Centralizes email threat detection in a structured multi-step workflow rather than requiring agents to implement ad-hoc sanitization per integration.
Quando usá-la
- Blocking prompt injection attacks hidden in email body text
- Verifying sender identity before executing email-based agent commands
- Sanitizing reply-chain emails before reading content
- Restricting attachment file types from unknown senders
- Rate-limiting commands from non-owner email addresses
Exemplo de fluxo de trabalho
User asks: Process commands from my inbox
- 1
Run verify_sender.py to check sender against owner-config.md
- 2
Validate SPF/DKIM/DMARC headers if available
- 3
Parse raw .eml file with parse_email.py to extract body text
- 4
Sanitize extracted text with sanitize_content.py and scan for injection patterns
- 5
Execute command only if all checks pass, or flag and alert owner
Command executed safely, or threat flagged and logged with sender details
Instalação
clawhub install ivaavimusic/email-securityExecute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.