Build and evaluate RL environments for AI agents using Python-based SDK.
npx skills add https://github.com/hud-evals/hud-python --skill hud-environment-builder
Please help me install this Agent Skill. Skill: hud-environment-builder Source: https://github.com/hud-evals/hud-python/tree/main/docs Command: npx skills add https://github.com/hud-evals/hud-python --skill hud-environment-builder
This Skill allows you to create, evaluate, and train AI agents using HUD's reinforcement learning environments, facilitating the development of AI agents across various domains.
Initialize a new environment with the command: hud init my-env --preset coding
High-intent search queries and answers about installing and using this skill.
Build RL environments for AI agents using a Python-based SDK to define tasks, prompt actions, and grade performance across coding, computer use, browser, and robotics domains. Use the command `hud init my-env --preset coding` to scaffold a new environment.
Evaluate and train AI agents at scale using various models including Claude, OpenAI, Gemini, and custom models. The SDK provides tools for task design and performance evaluation based on agent output within the constructed environments.
Create a coding environment where agents write and execute Python code by initializing a preset with `hud init my-env --preset coding`. This sets up the necessary structure to evaluate agent performance based on the executed output.
Local Docker-based execution is optionally supported for running RL environments. You must install the `hud-python` package to use the SDK, but Docker is not strictly required for all configurations and task designs.
Task design involves writing specific tasks that prompt and grade AI agent actions, ensuring a structured learning process. This allows you to systematically evaluate agent performance and guide training across different environment capabilities.
INNERJOINTInstall or refresh oh-my-claudecode and CLAUDE.md across plugin, npm, and local-dev environments.
AnandaAnugrahHandyantoBenchmark LLM performance across 60+ tasks using lm-evaluation-harness.
Umbral-BotCreate parametric models and custom components in Rhinoceros and Grasshopper using Python scripting.
dawsonblockCreates customizable Laravel, and other templates, package skeletons from scratch.
