The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands to perform environment pre-checks, such as verifying the installation of the futu-api SDK and ensuring the Futu OpenD client is running and reachable on the local machine.
[DATA_EXFILTRATION]: Transaction-related scripts automatically log order details to a local file (~/.futu_trade_audit.jsonl) to provide a persistent audit trail for the user's trading activities. This data remains local and is not transmitted to external third-party domains.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data including news headlines, company profiles, and analyst consensus from the Futu API. It mitigates potential injection risks by utilizing structured JSON outputs and providing predefined analysis templates in references/analysis-frameworks.md to prevent the direct reflection of raw external strings into the agent's reasoning logic.
[SAFE]: The skill implements significant safety guardrails for financial operations: it defaults to simulation mode (TrdEnv.SIMULATE), requires a mandatory --confirmed flag for real trades, and utilizes AskUserQuestion to ensure human verification before any order is placed.
[SAFE]: Programmatic unlocking of trading passwords via the unlock_trade API is strictly prohibited and explicitly refused by the skill instructions, forcing users to perform this action manually within the secure OpenD GUI for their protection.

futuapi