NetHack Agent Skills

Provide Python skills for AI agents to play NetHack via NetHackAPI.

13|1|Updated Jan 23, 2026

Search Tags:#automation #python #ai agent #rl #nethack #gameplay skills

Ready to run

npx skills add https://github.com/kenforthewin/glyphbox --skill nethack-agent-skills

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: NetHack Agent Skills
Source: https://github.com/kenforthewin/glyphbox/tree/main/skills
Command: npx skills add https://github.com/kenforthewin/glyphbox --skill nethack-agent-skills

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill Unit provides a structured framework and pre-built behaviors for an AI agent to play the complex game of NetHack, automating tasks like exploration, combat, and resource management.

Core Features & Use Cases

Modular Skills: Encapsulates specific game actions (e.g., eating, fighting, moving) into reusable Python functions.
AI Agent Integration: Designed to be called by an AI agent, allowing it to dynamically choose and execute skills based on game state.
Use Case: An AI agent can use the eat_when_hungry skill to ensure the player character doesn't starve, or the melee_fight skill to engage nearby monsters.

Quick Start

Execute the eat_when_hungry skill with the default hunger threshold.

Frequently Asked Questions about NetHack Agent Skills

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

FAQPage Schema

How do I automate NetHack gameplay with an AI agent in Python?▼

You can automate NetHack gameplay by using Python-based skills that handle core mechanics like combat, resource management, and navigation for an AI agent. These modular functions allow autonomous play based on game state queries.

What is needed to integrate AI agent skills for NetHack automation?▼

Integrating NetHack automation skills requires a NetHackAPI interface to query game states and execute actions. The Python functions rely on this interface to dynamically choose behaviors like eating when hungry or engaging in melee fights.

Can I use Python to manage combat and resource mechanics in NetHack?▼

Yes, you can use Python to manage combat and resource mechanics by calling modular skills like melee_fight and eat_when_hungry. These functions encapsulate specific game actions for an AI agent to execute during autonomous play.

How do I trigger autonomous eating in NetHack based on hunger levels?▼

You trigger autonomous eating by executing the eat_when_hungry skill with a default hunger threshold. This Python function ensures the AI agent manages resources effectively by preventing the player character from starving.

Does NetHack automation require reinforcement learning to play autonomously?▼

NetHack automation does not strictly require reinforcement learning, as the modular Python skills provide pre-built behaviors for exploration and combat. However, the framework supports integration with AI agents utilizing RL for dynamic skill selection.

What are the limitations of using modular Python skills for NetHack AI agents?▼

Limitations include dependency on a NetHackAPI interface for game state queries and action execution. The skills focus on core mechanics like navigation and combat, requiring an external AI agent to dynamically choose and execute them.