The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes documentation for running dig commands to verify DNS records such as SPF, DKIM, and DMARC. These are standard network diagnostic utilities and are used here in a safe, instructional context with no suspicious arguments or automated execution.
[DATA_EXFILTRATION]: No data exfiltration patterns detected. The skill references well-known services and organizations for email health monitoring (Google, Microsoft, Yahoo, Spamhaus, MXToolbox) and points to a vendor-related infrastructure resource (molted.email) in a neutral, informative manner.
[PROMPT_INJECTION]: No evidence of malicious prompt injection. The skill includes a dedicated section on 'AI agent email problems' that explicitly warns against the risk of indirect prompt injection from inbound email content and recommends loading security-focused skills to implement sanitization and detection guardrails.
[INDIRECT_PROMPT_INJECTION_SURFACE]: The skill identifies an attack surface regarding the processing of inbound email data.
Ingestion points: Inbound email content parsing is mentioned in the 'Setting up inbound email' section.
Boundary markers: The instructions recommend using delimiters and safety skills to handle untrusted content.
Capability inventory: The skill describes capabilities for DNS lookups and suggests integration with email sending/receiving APIs.
Sanitization: Explicitly mentions the need for HTML sanitization and Unicode character stripping before processing email data.

email-diagnostics