Apache-2.0 LLM multi-agent framework for simulating interactive, evolving story worlds; connects to DeepSeek Harness via the dsh-openstory plugin so agents can start simulations, direct characters and advance stories turn by turn.
DSH integration
Compatible
Author-claimed
Safety audit
Unaudited
Last verified
2026-08-21
License
Apache-2.0
01What can it help you accomplish?
Operate OpenStory multi-agent story simulations directly from DeepSeek Harness
Agent-driven simulation control via dsh-openstory: start simulations, view characters, issue instructions and advance the story turn by turn
DeepSeek Harness users who want interactive, evolving multi-agent story worlds controlled through their agent
Generate an interactive story world from a single sentence
WorldKernel turns one natural-language idea into characters, locations, paths and rules wired into a runnable multi-agent simulation
Story designers and hobbyists prototyping evolving story worlds without hand-authoring agents
Run and observe character simulations on a visual map
1:1 visual frontend with agent trajectories on the map, character profiles, real-time state changes and interaction archives; tick-by-tick simulation served by a local API server
Players and researchers exploring emergent social behavior in LLM agent communities
02How to install into DeepSeek Harness
Prerequisites
- Python 3.10 or above (recommended)
- Local Redis service started and listening on port 6379 (default data bus and cache)
- OpenAI-compatible model configuration for Westworld (examples/WestWorld/configs/models_config.yaml) and WorldKernel (examples/WorldKernel/configs/models.yaml or WORLDKERNEL_API_KEY env var)
Installation steps
- 01
Clone and install Agent-Kernel: `git clone https://github.com/ZJU-LLMs/Agent-Kernel.git && cd Agent-Kernel && pip install -e "packages/agentkernel-distributed[all]"`
$ git clone https://github.com/ZJU-LLMs/Agent-Kernel.git && cd Agent-Kernel && pip install -e "packages/agentkernel-distributed[all]"
- 02
Run the chosen story from the project root, e.g. Dream of the Red Chamber: `python -m examples.story_of_the_stone.run_simulation` (Westworld: `python -m examples.WestWorld.run_all`; WorldKernel: `pip install -e "examples/WorldKernel"` then `python -m worldkernel.server`)
- 03
For DeepSeek Harness, follow the dsh-openstory plugin configuration and usage guide linked in the README (dsh-openstory/README.zh.md)
Verify the integration
- Terminal prints `API Server started at http://0.0.0.0:8000`, then open http://localhost:8000/frontend/index.html in a browser (WorldKernel UI is at http://localhost:8100/)
03DSH integration and capability boundaries
Connects to DeepSeek Harness via the `dsh-openstory` plugin, letting the agent start simulations, view characters, issue instructions and advance the story turn by turn
Dynamic Agent-Kernel simulation
story examples, character data and YAML configs→multi-agent simulation that supports adding and removing agents dynamically during runtime
One-sentence world generation (WorldKernel)
one natural-language sentence describing a world idea→characters, locations, paths and rules generated into an interactive multi-agent simulation
generates local world data that can be continued latercalls an OpenAI-compatible model endpoint and requires local RedisVisual interactive frontend
a running simulation (local API server)→1:1 map view with agent trajectories, character profiles, state changes and interaction archives at http://localhost:8000/frontend/index.html
starts an API server listening on 0.0.0.0:8000 (WorldKernel uses port 8100)Plugin-based lifecycle & YAML configuration
YAML files under configs/ (simulation_config.yaml, models_config.yaml, system_config.yaml)→customized simulation rules: Pod count, max ticks, LLM model interface and parameters, Messager (message bus) and Timer (clock)
04Who is it for? When not to use it?
Good for
- DeepSeek Harness users who want interactive, evolving multi-agent story worlds controlled through their agent
- Story designers and hobbyists prototyping evolving story worlds without hand-authoring agents
- Players and researchers exploring emergent social behavior in LLM agent communities
Not for
- Requires a local Redis service running and listening on port 6379 as the default data bus and cache; entering WorldKernel simulations also depends on local Redis.
- Westworld and WorldKernel require you to fill in an OpenAI-compatible model configuration (models_config.yaml / models.yaml, or the WORLDKERNEL_API_KEY env var), so model calls go to an external endpoint you configure.
05Compatibility, maintenance and safety notes
- Requires a local Redis service running and listening on port 6379 as the default data bus and cache; entering WorldKernel simulations also depends on local Redis.
- Dream of the Red Chamber and Westworld both bind port 8000 and must not run simultaneously; WorldKernel uses port 8100 and can be started separately.
- Westworld and WorldKernel require you to fill in an OpenAI-compatible model configuration (models_config.yaml / models.yaml, or the WORLDKERNEL_API_KEY env var), so model calls go to an external endpoint you configure.
Apache-2.0 · actively maintained (last push 2026-08-19, release story-of-the-stone-zh-v1.0.0)
06Frequently asked questions
How do I connect OpenStory to DeepSeek Harness?
Through the dsh-openstory plugin. The README's news section links to the DSH plugin configuration and usage guide (dsh-openstory/README.zh.md); once connected, the agent can start simulations, view characters, issue instructions and advance the story turn by turn.
Is it a native runtime or MCP?
The README only mentions connecting OpenStory as a DeepSeek Harness plugin (dsh-openstory); MCP integration is not mentioned.
What prerequisites do I need?
Python 3.10 or above is recommended, and a local Redis service must be running on port 6379. Westworld and WorldKernel additionally require an OpenAI-compatible model configuration (models_config.yaml / models.yaml, or the WORLDKERNEL_API_KEY env var).
Which ports does it use?
Dream of the Red Chamber and Westworld both use port 8000, so they cannot run at the same time. WorldKernel uses port 8100 and can be started separately.
Where do my data and network requests go?
Local Redis acts as the default data bus and cache; model calls go to the OpenAI-compatible endpoint you configure yourself. The frontend is served locally at localhost:8000 (WorldKernel at 8100).
07Related DSH workflows
openviking
by volcengine
Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
colleague-skill
by titanwings
Turn memories and knowledge into warm, reusable agent skills — a skill generator for digital companions in the AI persona era.
archify
by tt-a1i
Agent skill for beautiful, verifiable architecture, workflow, sequence, data-flow, and lifecycle diagrams—self-contained HTML with motion and crisp export.
learn-harness-engineering
by walkinglabs
Harness engineering beginner tutorial, from 0 to 1
08Data and sources
**✨ OpenStory 现已支持 DeepSeek Harness 插件！** 现在可以通过 `dsh-openstory` 将 OpenStory 多智能体推演接入 DeepSeek Harness，让 agent 直接启动模拟、查看…
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-08-21. Found an error? Submit a correction.