Archie Personal Assistant

This is an exploratory project to create a personal assistant service which leverages a locally running llm server such as Ollama (https://github.com/ollama/ollama), supporting models with tool and reasoning capabilities.

Note

This is an experimental project in which big changes will inevitably occur.

Requirements

A locally running LLM server such as Ollama installed with the required model pulled.

The use of a model with tools and thinking capabilities.

For a list of models that support these capabilities, refer to https://ollama.com/library

Architecture

Tools can be added in the form of class libraries which need to be present in a directory defined in archie-service.config.json. By default, it will search the current directory recursively.

These will be loaded and tool classes registered at runtime.

Console Interface

Provided is a console interface project which provides a basic interface to communicating with the agent.

Screenshots