The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of local command execution to interact with the system.
Uses osascript to execute AppleScript for controlling Mail.app functionality (sending, reading, and managing messages).
Uses sqlite3 to query the local Apple Mail database (~/Library/Mail/V{9,10,11}/MailData/Envelope Index) for fast metadata retrieval.
Uses python3 for parsing email files (.emlx) and URL decoding.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the way it processes external data.
Ingestion points: Incoming emails read via mail-read.sh, mail-list.sh, and mail-read-emlx.py provide untrusted content to the agent.
Boundary markers: The output format does not include explicit delimiters or warnings to the agent to ignore instructions embedded within the email bodies.
Capability inventory: The skill provides significant 'write' capabilities, including mail-send.sh, mail-reply.sh, and mail-delete.sh.
Sanitization: Scripts include basic escaping for shell/AppleScript command construction, but there is no content filtering or sanitization to prevent an agent from obeying instructions found inside a processed email body.

apple-mail