The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the agentmail CLI tool to perform various email-related tasks such as creating inboxes, sending messages, and managing webhooks.
[EXTERNAL_DOWNLOADS]: Instructions are provided to install the agentmail-cli package from npm, which is the standard distribution method for this utility.
[CREDENTIALS_UNSAFE]: The skill requires an AGENTMAIL_API_KEY. It correctly advises users to provide this via environment variables or command flags, avoiding hardcoded secrets.
[INDIRECT_PROMPT_INJECTION]: The skill functions by reading incoming emails, which constitutes a surface for indirect prompt injection from untrusted external content.
Ingestion points: Commands such as agentmail inboxes:messages retrieve in SKILL.md read message bodies into the agent context.
Boundary markers: No specific delimiters or safety warnings for the agent are defined in the instructions.
Capability inventory: The agent can execute agentmail commands to send emails, forward threads, or create webhooks.
Sanitization: No explicit content sanitization or validation logic is specified in the skill configuration.

agentmail-cli