#6363rd of 70 in Agent frameworks and SDKs
upsonic
Python framework for building autonomous and tool-calling AI agents
- Stars
- 8k
- License
- MIT
- Last commit
- Jun 2026
- Last release
- May 2026
- Language
- Python
7,744 on 1 Apr 20267,962 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Upsonic is a Python library with two agent types: an AutonomousAgent that works inside a restricted workspace directory (file and shell operations confined, path traversal and dangerous commands blocked), and a traditional Agent with custom tools via a @tool decorator and MCP support. It also ships a layered OCR pipeline supporting EasyOCR, RapidOCR, Tesseract, PaddleOCR and DeepSeek OCR.
Who it is for: Python developers building agents that use tools, MCP and file or shell access
Strengths
- Autonomous agent confines file and shell operations to a configured workspace
- Plain Python tools via @tool decorator, plus MCP tool integration
- Built-in OCR with six engines, including local ones like Tesseract and EasyOCR
- Community-contributed prebuilt agents bundle a skill, system prompt and first message
Weaknesses
- README examples only show Anthropic model strings; other provider support is not listed
- Isolated cloud execution needs an external sandbox provider (E2B)
- No hardware requirements or deployment guidance in the README
- Workspace restriction is described briefly; no detail on how dangerous commands are detected
What it needs
- Needs E2B (optional sandbox), LLM provider API (e.g. Anthropic)
- Models: anthropic/claude-sonnet-4-5, DeepSeek OCR via Ollama
Also in Agent frameworks and SDKs
See all 70| Rank | Project | Score |
|---|---|---|
| 1 | ppt-masterAgent skill that turns documents into natively editable PowerPoint decks | 94 out of 100 |
| 2 | langchainPython framework for building LLM agents and apps from interoperable components | 89 out of 100 |
| 3 | browser-usePython library that lets an LLM agent control a web browser | 89 out of 100 |
| #4 | openai-agents-pythonPython SDK for multi-agent workflows with handoffs, guardrails and tracing | 89 out of 100 |
| #5 | mastraTypeScript framework for building AI agents and workflows | 89 out of 100 |