STORM wiki

STORM frontend modified.

Features & Changes

• dotenv for env vars
• save result files as '*.md'
• add date to to top of the result file
• added STORM_TIMEZONE
• configure dark and light theme colors
• configure arize-phoenix in settings
• configure searxng and other search engines in settings

screenshots

article

creating new article with search and llm fallback options

add categories

configure search engines

dark and light themes

Prerequisites

• Docker and Docker Compose
• Required API keys (see main STORM repository)

Installation and Usage with Docker

1. Clone the repository:

   git clone https://github.com/jaigouk/storm_wiki.git
   cd storm_wiki

2. Set up environment variables:

   cp .env.example .env
   # Edit .env with your preferred text editor to set the required variables

3. Set up Streamlit secrets:

   cp secrets.toml.example secrets.toml
   # Edit secrets.toml with your preferred text editor to set the required secrets

4. Build and run the Docker container:

   docker compose up --build

   This command will build the Docker image and start the container. The app will be available at http://localhost:8501 (or the port you specified in the .env file). If you want to use specific directory then change STREAMLIT_OUTPUT_DIR.

5. To stop the container:

   docker compose down

Configuration

After starting the app, you can configure various settings through the UI:

1. Configure search engines:

   • Visit the settings menu and click Search settings
   • Add necessary values. Note: Bing and You search are currently not usable, but you can use searxng, arxiv, duckduckgo
   • Choose primary and fallback search engines

2. Configure LLM:

   • Visit the LLM menu and choose primary and fallback LLM
   • For ollama, you will see the localhost LLM list. Choose them with max tokens
   • You can set a fallback LLM like OpenAI's gpt-4o-mini

3. Categories setting:

   • Choose output directory (root directory for category folders)
   • Edit and delete existing categories
   • Create new category folders

Customization

To customize STORMWikiRunner settings, modify set_storm_runner() in demo_util.py. Refer to the main STORM repository for detailed customization options.

Development

If you want to run the app locally for development:

1. Install Python 3.10+
2. Install dependencies:

   pip install -r requirements.txt

3. Run the Streamlit app:

   streamlit run storm.py --server.port 8501 --server.address 0.0.0.0

Troubleshooting

If you encounter any issues with the Docker setup, ensure that:

• Docker and Docker Compose are installed and up to date
• The required ports are not being used by other applications
• Your .env and secrets.toml files are properly configured

For more detailed logs, you can run:

docker compose up --build --log-level DEBUG