Awesome MCP › Agent Orchestration Platforms

vortezwohl/Autono

⭐ 212 Python added to this list on 2025-04-09 repository created 2024-10-25

Autono is a highly robust autonomous agent framework based on the ReAct paradigm, designed to enhance adaptive decision making and multi-agent collaboration for complex task execution. Unlike traditional frameworks that rely on fixed workflows generated by large language model (LLM) planners, Autono dynamically generates next actions during agent execution based on prior trajectories, improving robustness and flexibility. It introduces a timely abandonment strategy with a probabilistic penalty mechanism to address potential termination issues caused by adaptive execution paths. This strategy allows developers to balance conservative and exploratory behaviors by tuning hyperparameters, significantly improving adaptability and efficiency in complex environments. The framework supports multi-agent collaboration through a memory transfer mechanism that enables shared and dynamically updated memory among agents. This explicit division of labor allows agents to focus on specific task components, enhancing execution efficiency and quality. Autono is modular and supports external tool integration, with compatibility for the Model Context Protocol (MCP), enabling flexible expansion of the agent's action space. Experimental results demonstrate that Autono outperforms other frameworks like Autogen and Langchain in handling tasks of varying complexity, especially multi-step tasks with possible failures. It supports various models including gpt-4o-mini, qwen-plus, and deepseek-v3, achieving high success rates in one-step, multi-step, and failure-prone multi-step tasks. The framework is easy to install via PyPI or GitHub and provides a quick start guide for building agents with customizable personalities and abilities. Users can define abilities as functions, instantiate agents with these abilities, and assign tasks for autonomous execution. The framework encourages citation in academic work and offers multilingual documentation. Keywords: autonomous agent, ReAct paradigm, adaptive decision making, multi-agent collaboration, Model Context Protocol, MCP, timely abandonment strategy, probabilistic penalty, memory transfer, modular design, external tool integration, task execution, LLM, agent framework, multi-step tasks, robustness, agent personality, ability management.

https://github.com/vortezwohl/Autono

abilitiesability-managementadaptive-decision-makingagentagent-frameworkagent-personalitiesagent-personalityagentic-abilitiesagiaiaiagentautogenautonomous-agentautonomous-agentsexternal-tool-integrationframeworklangchainllmllm-frameworkmcpmcp-agentmcp-agent-frameworkmcp-clientmemory-transfermemory-transfer-mechanismmodel-context-protocolmodular-designmulti-agent-collaborationmulti-step-tasksmultiagentnlpopenaiopenai-modelprobabilistic-penaltypythonreactreact-paradigmrobust-frameworkrobustnesstask-executiontask-execution-efficiencytimely-abandonment-strategytool-learningtool-oriented-learningtransformer