<div align="center"><img src="https://github.com/user-attachments/assets/211d0c2b-04de-471e-b1ed-97da94a58d82" height="20"/></div>
Free[GPT Computer Assistant](https://github.com/Upsonic/gpt-computer-assistant)
FreeFree tier
About <div align="center"><img src="https://github.com/user-attachments/assets/211d0c2b-04de-471e-b1ed-97da94a58d82" height="20"/></div>
Upsonic is an open-source Python framework for building autonomous AI agents, including prebuilt agents like OpenClaw and Claude Cowork, as well as traditional agent systems. It supports multiple AI models (e.g., anthropic/claude-sonnet-4-5), workspace-restricted execution for safety, custom tools via decorators, MCP tools integration, and a layered OCR pipeline for document processing. Installation is via pip, and the framework offers IDE integration with Cursor, VSCode, and Windsurf. It is designed for developers looking to create intelligent automation systems with Python.
Key Features
Build autonomous agents and traditional agent systems with Python
Supports various AI models (e.g., Claude, OpenAI) via provider/model strings
Workspace-restricted execution for safety – blocks path traversal and dangerous commands
Custom tool creation using @tool decorator
Collection of prebuilt autonomous agents ready to run
MCP tools integration for connecting to external data sources and services
OCR and document processing pipeline with layered architecture (Layer 0: document prep, Layer 1: OCR engine)
IDE integration by adding Upsonic docs to Cursor, VSCode, Windsurf
Sandbox provider (E2B) support for isolated cloud execution
Pros & Cons
Pros
- Open source and free to use
- Easy installation via pip
- Flexible: supports both autonomous and traditional agent paradigms
- Safety features with workspace restrictions prevent unauthorized access
- Extensible with custom Python tools and MCP integrations
- Prebuilt agents available for quick start
- Built-in OCR pipeline for document processing tasks
Cons
- Requires Python programming knowledge, not suitable for non-developers
- May need a cloud sandbox provider (e.g., E2B) for full isolated execution
- Documentation may still be evolving (currently only docs link provided)
- Relatively new project; ecosystem and community are still growing
Best For
Analyze server logs and detect anomaly patterns autonomouslyAnalyze current market trends with traditional agentsPerform calculations using custom tools (e.g., sum of numbers)Document processing and OCR extractionBuild AI assistants for automation and data analysis tasksIntegrate with external tools and data via MCP
FAQ
What is Upsonic?
Upsonic is a Python framework for building autonomous AI agents, including prebuilt agents like OpenClaw and Claude Cowork, as well as more traditional agent systems.
How do I install Upsonic?
Install with pip: pip install upsonic. For OCR support, use pip install 'upsonic[ocr]'.
What models can I use with Upsonic?
You can specify any model, e.g., 'anthropic/claude-sonnet-4-5'. It works with various providers and models.
Is Upsonic safe to use?
Yes, all file and shell operations are restricted to the workspace, and path traversal and dangerous commands are blocked.
Can I create custom tools?
Yes, define tools with the @tool decorator and pass them to tasks. You can create custom Python functions as tools.
Does Upsonic support OCR?
Yes, it provides a unified OCR interface with a layered pipeline: Layer 0 for document preparation and Layer 1 for the OCR engine (e.g., EasyOCR).