The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses bash to perform a syntax check on generated Python code via python -c "import ast; ast.parse(...)". This is a standard and safe developer utility used for validating code structure without executing its logic.
[DATA_EXPOSURE]: The skill involves reading market data and writing strategy configurations and code to a local directory. It does not access sensitive system paths, credentials, or environment variables.
[REMOTE_CODE_EXECUTION]: While the skill generates Python code (signal_engine.py), the execution is performed by a built-in backtest tool within the agent's environment. The instructions explicitly forbid external libraries and complex executions, keeping the logic within a sandboxed pandas/numpy context.
[PROMPT_INJECTION]: The instructions contain standard directives for parsing user intent and normalizing inputs. There are no attempts to bypass safety filters or override agent constraints.
[EXTERNAL_DOWNLOADS]: The skill documentation references data sources like tushare, yfinance, and okx. These are well-known financial data providers and their use is consistent with the primary purpose of a trading strategy skill.

strategy-generate