The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands via the longbridge CLI to manage user price alerts. These operations are scoped to price alert management and do not involve financial transactions or broader system access.
[SAFE]: The skill enforces a mandatory 'Two-step protocol' for all mutating operations (add, delete, enable, disable). The agent must provide a preview of the action and wait for explicit user confirmation before executing the command.
[SAFE]: The skill correctly directs users to the official vendor repository on GitHub (github.com/longportapp/longbridge-terminal) for the required CLI tool, ensuring users obtain software from a legitimate source.
[SAFE]: Authentication is handled via the official longbridge auth login command, leveraging a secure OAuth flow rather than asking for or storing raw credentials within the skill or prompt.

longbridge-alert