CaseLens Agents

Agents for the CaseLens project. Built with BESSER Agentic Framework

Run the app

Requirements

• Python >= 3.10
• Recommended: Create a virtual environment (e.g. venv, conda)

pip install -r requirements.txt

python run.py

You can access the application in http://localhost:8501

Deploy with Docker

1. Build Docker image

This command uses the file Dockerfile

docker build -t besser-agents .

2. Create and run container

This command uses the file docker-compose.yml

docker-compose up -d

You can access the application in http://localhost:8501

Volumes

In Docker, a volume is a persistent storage mechanism used to store and share data between containers and the host system, independent of the container's lifecycle.

This app has 2 volumes, as specified in docker-compose.yml:

• .streamlit: contains secrets.toml and config.toml, which store some Streamlit properties (Streamlit is the UI framework of the application) (more info in the official Streamlit docs)
• data: Stores data created by the agents.

The data volume contains the following:

• config.ini: properties for the agents. We can define the following properties here:
  • nlp.ollama.host = localhost Host address of the Ollama LLM
  • nlp.ollama.port = 11434 Port of the Ollama LLM
  • nlp.ollama.max_tokens = 8000 Maximum number of input tokens for the LLM
  • nlp.ollama.model = gemma3:12b Name of the Ollama LLM (full list here)
  • nlp.hf.tokenizer = google/gemma-2-2b-it Name of the tokenizer to use (should be the same family of the LLM. (full list here))
  • nlp.hf.api_key = YOUR-API-KEY HuggingFace API Key. Some tokenizers may need authentication and therefore it is necessary to provide this key.
  • elasticsearch.host = localhost Host address of the elasticsearch database
  • elasticsearch.port = 19200 Port of the elasticsearch database
  • elasticsearch.index = castor-test-enron Name of the elastiscearch index
• data_labeling_agent folder: Contains the file request_history.json, which stores the requests done with this agent.
• chat_files_agent folder: Contains the file chat_notebook.json, which stores the requests done with this agent. Also contains the chats folder. All imported chats are processed and exported in JSON format into this folder. The agent actually uses these files to analyze the chat files.