The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and installs the wqo==0.1.1 package from the official PyPI registry (https://pypi.org/simple). This is performed by the scripts/wqo.sh script using either uv or pip to set up a local virtual environment within the user's project directory.
[COMMAND_EXECUTION]: The scripts/wqo.sh script uses standard shell commands to manage environment setup and execute the research CLI. It utilizes exec with preserved arguments ("$@") to run the wqo binary, which is a safe practice for command forwarding.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it processes research alpha expressions and data identifiers provided in the conversation.
Ingestion points: Research code strings passed to the wqo sim run --code flag and dataset/alpha IDs used across various commands.
Boundary markers: The references/runtime.md file provides extensive operational boundaries, including explicit instructions for the agent to never automate submissions without user confirmation and to avoid accessing credential files directly.
Capability inventory: The skill can perform network operations via the installed WQO CLI, execute shell commands via scripts/wqo.sh, and write to the local .wqo/ project subdirectory.
Sanitization: The runner script employs umask 077 to restrict file permissions and set -euo pipefail for execution safety. The skill relies on the WorldQuant BRAIN API and the CLI tool's own validation for sanitizing simulation code.

wqo-cli