# Eva01

An experimental autonomous AI agent with a multimodal cognitive architecture, built-in voice and face recognition, and pluggable tools, designed to act on its own initiative rather than only respond to commands.

Eva01 is an autonomous agent that runs continuously as a concurrent pipeline: senses feed a shared buffer, a reasoning graph consumes and processes them, and actions flow out through actor modules. Its cognitive architecture is modeled in three layers running at different speeds: an autonomic layer that checks the system's own health, a subconscious layer that gates which perceptions reach reasoning, and a conscious reasoning graph above them, intended to decide when a model call is actually needed rather than running on a fixed clocked heartbeat.

It includes built-in voice and face recognition, is multimodal, and is compatible with ChatGPT, Claude, DeepSeek, Gemini, Grok and Ollama as underlying models, with a modular, plug-and-play tool system.

It is written in Python and licensed under MIT.

Website: https://genesis1231.github.io/Eva01/journal