The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation process for the required CLI tool involves downloading shell and PowerShell scripts from a remote GitHub repository (FutunnOpen/futu-cli) and executing them immediately. On macOS and Linux, this is done by piping a remote script directly into the shell (curl | sh). On Windows, it downloads a .ps1 file and executes it.
[PRIVILEGE_ESCALATION]: To install the tool on Windows, the skill instructs the agent to run PowerShell with the -ExecutionPolicy Bypass flag. This deliberately circumvents local security policies that restrict the execution of scripts, potentially allowing unsigned or untrusted code to run.
[COMMAND_EXECUTION]: The skill relies on the Bash tool to run the futu CLI for all operations. This includes probing for the executable, checking authentication status, and executing financial trades. The agent is instructed to use these commands to manage accounts, funds, and positions.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes data from external command outputs. 1. Ingestion points: CLI command outputs (standard output and JSON) containing account details, order remarks, and transaction logs. 2. Boundary markers: The skill requires the use of --format json to delimit output and provide structure, but does not explicitly instruct the agent to ignore instructions embedded in those fields. 3. Capability inventory: The skill uses the Bash tool to execute a wide range of trading commands, including placing, modifying, and canceling orders. 4. Sanitization: The skill relies on smallest parameter set generation and explicit user confirmation for asset-changing commands to mitigate risk.

futu-trading