**Language:** [![English](https://img.shields.io/badge/English-English-blue?style=flat-square)](README.en.md) [![中文](https://img.shields.io/badge/%E4%B8%AD%E6%96%87-%E4%B8%AD%E6%96%87%E7%89%88-blue?style=flat-square)](README.md)

# UNIVERSAAL Agent

[![Python](https://img.shields.io/badge/Python-3.10-%233776AB?logo=python)]()
[![Status](https://img.shields.io/badge/Status-Planning-blue)]()
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)

An experimentation and planning repository for a general-purpose agent system, exploring a unified architecture for cross-domain task planning, tool integration, and knowledge management.

## Status

> **Planning**: currently only design ideas and initial notes exist, with no complete implementation yet — happy to be filled in step by step later.

## Research Directions

| Topic | Description |
|-------|-------------|
| Cross-Domain Task Planning | Unified decomposition and scheduling methods for tasks across domains |
| Tool Integration | Registration, invocation, and error handling of diverse external capabilities |
| Knowledge Management | Long-term memory, semantic retrieval, and knowledge-base update strategies |
| Interaction Design | Design of human-agent collaboration interfaces and feedback mechanisms |

## Planned Structure

```
UNIVERSAAL_AGENT/
├── docs/           # design documents and research notes
├── experiments/    # proof-of-concepts and experiments
├── core/           # core framework prototype
└── README.md
```

## Related Projects

This repository belongs to the Agent research series; the division of labor among the three:

| Repository | Role |
| --- | --- |
| [Modern_agent](https://github.com/TeWei02/Modern_agent) | Framework-layer design: tool use, context memory management, multi-model integration |
| [UNIVERSAAL_AGENT](https://github.com/TeWei02/UNIVERSAAL_AGENT) | General-purpose architecture: cross-domain task planning, tool integration, knowledge management |
| [agent_full](https://github.com/TeWei02/agent_full) | Hands-on experiments: implementation records for macOS local models and RAG |

## License

MIT © [Te-Wei Ko](https://github.com/TeWei02)