Story Writing

App Information

Using the website

To use the website, all you need to do is use streamlit to run the app locally. First, clone the repository and then once you are in the repository directory call `streamlit run App_Information.py`. If you are hosting your own graph (see [this documentation](https://langchain-ai.github.io/langgraph/cloud/deployment/cloud/) for hosting on LangGraph Cloud), make sure to change the url in the `get_client` function call in `pages/Story_Writing.py` to match the url of your deployed graph.

Once you have called `streamlit run App_Information.py` you can go to `http://localhost:8501` to see the application.

If you don't wish to host yourself, you can access the website by going to the following URL: https://langchain-story-writing.streamlit.app.

Background

This app was designed to show off some LangGraph Cloud features in a fun, interactive way. This app is designed to allow users to write a story with the help of a LangGraph agent. The app allows users to edit chapters they have written already, or continue the story by writing the next chapter. This means the user can have multiple versions of the same chapter number and can select the one they like most to continue the story from. At the beginning the user provides the graph information on the summary of the story, the writing style they want, and any additional details important to the story. From that point they just need to provide edit and continue instructions to steer the agent in the desired direction.

*Note: This app is a prototype and not ready for deployment. There might be bugs/poor results from the agent.*

The Graph State

One of the coolest features of LangGraph Cloud is the ability to have a persistent state across many runs of the graph. In this case we are able to retain information about the story as the graph continues to write it. In our case, we keep an overall chapter state graph, which is just a dictionary containing the different chapters writte