The Agent Skills Directory

[Data Exposure] (LOW): The script reads and writes sensitive authentication data, including OAuth 2.0 tokens and client secrets, to the ~/.contacts_credentials/ directory. While necessary for operation, these files should be protected with appropriate filesystem permissions.
[Indirect Prompt Injection] (LOW): The skill ingests untrusted data from the Google People API (contact names, notes, etc.). While the current implementation only prints this data to stdout, an attacker who can modify a user's contacts could theoretically place instructions in contact fields to influence the agent's behavior during subsequent processing.
[COMMAND_EXECUTION] (SAFE): Commands are executed via standard library arguments and do not involve shell interpolation of untrusted strings.

google-contacts