OpenStory is a multi-agent simulation framework leveraging Large Language Models (LLMs) and an Agent-Kernel architecture. It enables dynamic, AI-driven narrative deduction and world-building, targeting developers and researchers interested in interactive storytelling and complex AI agent interactions. The framework allows for creative co-creation of stories, offering a novel approach to AI-powered narrative experiences.

How It Works

OpenStory employs an Agent-Kernel architecture enabling dynamic agent addition/removal during simulation, offering high extensibility. LLMs drive agent behaviors and plot generation, fostering deep interactions based on personalities and context. A 1:1 simulation of "Dream of the Red Chamber's" Grand View Garden features a visual interface tracking agent movements and profiles, enhancing immersion.

Quick Start & Requirements

Installation involves cloning the Agent-Kernel repo, pip installing dependencies, and running the simulation.

• Primary install / run command:

  git clone https://github.com/ZJU-LLMs/Agent-Kernel.git
  cd Agent-Kernel
  pip install -e "packages/agentkernel-distributed[all]"
  cd ../../..
  python -m examples.story_of_the_stone.run_simulation
  

• Non-default prerequisites: Python 3.10+ and Redis (port 6379).
• Links: Interactive tutorial at http://localhost:8000/frontend/index.html (post-simulation).

Highlighted Details

• Dynamic Agent Framework: Agents can be added or removed during simulation runtime.
• Interactive Visualization: A 1:1 replica of the Grand View Garden with real-time agent tracking and detailed character profiles.
• AI-Generated Plotlines: System generates dramatic and engaging narratives based on agent interactions and environmental factors.
• Plugin System: Supports full agent lifecycle plugins and highly configurable via YAML files.

Maintenance & Community

The project is actively developing, with "Story 2" announced as "Coming Soon." A QQ community group (1091827223) is available for users. No specific details on core maintainers, sponsorships, or partnerships are provided.

Licensing & Compatibility

The project's license is not specified in the README. This absence is a critical omission for assessing commercial use or integration into closed-source projects.

Limitations & Caveats

The README does not detail known limitations, bugs, or unsupported platforms. The absence of license information is a significant caveat for adoption. Ongoing development is indicated by upcoming content modules.