The Agent Skills Directory

[COMMAND_EXECUTION]: The SKILL.md file instructs the agent to execute a shell command (ls) to check for the presence of recommended companion skills in various local directories such as ~/.claude/skills/. This discovery is used to offer relevant installation commands to the user.
[PROMPT_INJECTION]: The skill defines procedures for parsing external email data, specifically Delivery Status Notifications (DSN) and Abuse Reporting Format (ARF) files. This creates an indirect prompt injection surface as these inputs originate from external sources.
Ingestion points: Email headers and bounce report bodies described in references/bounce-handling.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are included in the parsing instructions.
Capability inventory: The agent possesses shell execution capabilities (used for ls and npx commands) and general file system access.
Sanitization: The skill lacks instructions for sanitizing or validating the content of external email reports before they are processed by the agent.
[SAFE]: All external URLs and technical references target reputable organizations and official documentation (IETF RFCs, Google Postmaster Tools, Microsoft SNDS, Spamhaus). The implementation of installation and discovery logic is consistent with the skill's stated purpose as a production-ready AI agent extension.

email-deliverability