The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/wqo.sh script automatically handles the installation of the wqo package from the official Python Package Index (PyPI). This installation is constrained to a local project virtual environment (.wqo/venv), preventing modifications to the global system or user environments.
[COMMAND_EXECUTION]: The skill executes the wqo CLI tool via a shell wrapper script. The wrapper uses umask 077 to ensure that all created files and directories, including the virtual environment, have restrictive permissions that prevent access by other users on the system.
[INDIRECT_PROMPT_INJECTION]: The skill ingests alpha expressions and JSON-formatted settings files which are then passed to the simulation tool.
Ingestion points: Simulation code passed via the --code flag and batch files processed via wqo sim batch.
Boundary markers: The instructions specify strict formatting requirements (no scientific notation, bare identifiers for groups) which act as a form of schema validation.
Capability inventory: The skill can perform shell executions through wqo.sh and network operations to the WorldQuant BRAIN API via the CLI.
Sanitization: The underlying CLI tool sanitizes inputs before transmission to the remote API, and the documentation provides clear rules for expression syntax.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill provides security-conscious guidance for credential management. It explicitly instructs the user to manually create a private credentials file and set restricted permissions (chmod 600), while strictly forbidding the agent from reading, writing, or logging the contents of that file.

wqo-simulate