The Agent Skills Directory

[SAFE]: The skill instructions explicitly prohibit the agent from requesting, accepting, or storing sensitive credentials such as API keys or OAuth tokens, ensuring that authentication remains handled by the secure client environment.
[SAFE]: A mandatory confirmation loop is enforced for all message-sending actions, requiring a clear payload preview and explicit user approval immediately before the tool is called to prevent unauthorized or accidental actions.
[SAFE]: Privacy controls are integrated into the instructions, requiring the agent to mask phone numbers and minimize the display of sensitive message content to reduce data exposure.
[SAFE]: Robust error-handling logic prevents automated retries for ambiguous operations, instead requiring manual verification of the message state to avoid accidental duplication.

sent-messaging