scienceworld-task-interpreter

Parse ScienceWorld task instructions to extract objectives and locations.

1.1k|131|Updated Jan 10, 2026
One-click install
npx skills add https://github.com/zjunlp/SkillNet --skill scienceworld-task-interpreter
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: scienceworld-task-interpreter
Source: https://github.com/zjunlp/SkillNet/tree/main/experiments/src/skills/scienceworld/scienceworld-task-interpreter
Command: npx skills add https://github.com/zjunlp/SkillNet --skill scienceworld-task-interpreter

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This skill simplifies complex user instructions in the ScienceWorld environment by breaking them down into actionable steps, ensuring the AI agent can efficiently navigate, observe, and interact with the virtual world.

Core Features & Use Cases

  • Task Parsing: Identifies the core objective, target, and location from user commands.
  • Plan Generation: Creates a step-by-step plan including navigation, observation, and execution.
  • Environment Interpretation: Understands ScienceWorld's specific locations and object interactions.
  • Use Case: When a user says, "Find the animal with the shortest life span in the outside area," this skill will interpret the objective (find shortest life span), target (animal), and location (outside), then generate a plan to teleport outside, look around, and identify the correct animal.

Quick Start

Use the scienceworld-task-interpreter skill to parse the instruction "find the red ball in the living room".

Frequently Asked Questions about scienceworld-task-interpreter

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I parse ScienceWorld instructions to extract task objectives and locations?

To parse ScienceWorld instructions, the task interpreter identifies core objectives, targets, and locations from user commands, outputting a clear, actionable sub-goal for agent execution.

How does task interpretation work for game AI agents in interactive environments?

Task interpretation for game AI agents works by breaking down complex user instructions into actionable steps, identifying the objective, target, and location to generate a navigation and execution plan.

Can I generate a step-by-step plan for finding specific objects in the ScienceWorld environment?

Yes, you can generate a step-by-step plan for finding objects. The skill parses the instruction to identify the target and location, then creates a plan involving navigation, observation, and execution.

Does the ScienceWorld task interpreter support environment interaction and navigation planning?

Yes, the ScienceWorld task interpreter supports environment interaction and navigation planning by understanding specific locations and object interactions to generate actionable sub-goals for agent execution.

What's the best way to break down complex scientific tasks into actionable sub-goals for an AI agent?

The best way to break down complex scientific tasks is by parsing high-level instructions to extract the objective, target, and location, simplifying them into a clear, actionable sub-goal for execution.