The Agent Skills Directory

[COMMAND_EXECUTION]: The skill orchestrates its multi-agent workflow by calling several local utility scripts (e.g., deep-research-bootstrap.sh and deep-research-deliver.sh) to manage workspace tasks and finalize deliverables.
[EXTERNAL_DOWNLOADS]: The fetch-fred command in scripts/research_quant_toolkit.py utilizes the requests library to download economic data from the official St. Louis Fed (FRED) CSV data service.
[REMOTE_CODE_EXECUTION]: The skill includes a calculation utility in scripts/research_quant_toolkit.py that uses eval() to evaluate user-supplied mathematical expressions. The implementation uses ast.parse() to strictly limit allowed operations to safe mathematical nodes and executes the code with no built-in functions available to prevent arbitrary code execution.
[SAFE]: Implements a structured research methodology with automated reference normalization and integrity checks (via deep_research_reference_guard.py) to ensure that all generated research findings are verifiable and derived from legitimate sources.

deep-research