The Agent Skills Directory

[COMMAND_EXECUTION]: The skill manages system-level scheduled tasks (cron jobs) to automate market scanning, position monitoring, and risk management. This includes instructions for the agent to create, enable, and disable cron jobs for specific assets.
[DATA_EXFILTRATION]: The skill is designed to send trade confirmations and status updates to external services via Telegram. This involves network operations to a third-party platform for operational notifications.
[SAFE]: The skill identifies and documents a critical platform-level bug where the dryRun flag is ignored by the create_position tool, which could lead to real-money execution. It provides specific instructions to avoid this tool parameter to ensure financial safety.
[SAFE]: The skill implements a robust 'Hourly Trend Alignment' rule and multiple risk-management filters (directional exposure guards, daily loss limits, drawdown caps) to ensure safe autonomous operation and prevent correlated portfolio blowups.
[INDIRECT_PROMPT_INJECTION]: The skill processes market data, smart money signals, and trader counts from external APIs (e.g., Hyperliquid) to drive trading decisions. While this data influences agent behavior, the skill applies strict logic-based filters such as trend alignment and score thresholds to mitigate potential manipulation from these external data sources.

autonomous-trading