Communitygithub.com

GTLordLL/CutinAgent

CutinAgent (Thousand Tasks Mini Cutter): An SOP-driven, controlled Agent framework built with Python & LangGraph. Runs LLMs as standardized digital executors on consumer-grade hardware (e.g., RTX 3060 6GB, qwen3:4b).

What is CutinAgent?

CutinAgent is a lightweight, controlled agent framework built with Python and LangGraph, designed to run LLMs as standardized digital executors on consumer-grade hardware like an RTX 3060 with 6GB VRAM (e.g., using qwen3:4b). Its core strength lies in its Standard Operating Procedure (SOP)-driven architecture: workflows are defined as strict, reproducible YAML-based SOPs, combining a trigger, an LLM executor, and a converter to transform outputs into structured data. This allows developers to build predictable, auditable AI agents that can be chained into complex multi-step tasks without relying on expensive cloud APIs. The framework supports platforms like LangGraph for graph-based orchestration and Python for extensibility, making it ideal for edge computing, local deployments, and cost-sensitive applications where reliability and repeatability are paramount.

Works with~Claude Code~Codex CLI~Cursor
npx skills add GTLordLL/CutinAgent

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

What does CutinAgent do?

CutinAgent (Thousand Tasks Mini Cutter): An SOP-driven, controlled Agent framework built with Python & LangGraph. Running LLMs as standardized digital executors on consumer-grade hardware (e.g., RTX 3060 6GB, qwen3:4b).

Related Skills