mini-agent-framework Alternatives

agent-mini is an open-source CLI tool that acts as a lightweight personal AI agent, automating shell, file, and web tasks. It supports both local and cloud LLMs, making it ideal for developers and power users seeking AI-driven automation from the command line.

agentos-framework is an open-source agent framework for developing, orchestrating, and managing autonomous AI agents. It supports multiple LLM providers, function calling, streaming, checkpointing, and swarm coordination, making it suitable for advanced AI development and research.

agent-framework is an open-source Python framework developed by Microsoft for building, managing, and deploying AI agents. It includes core and optional packages, supporting modular development for AI-powered applications.

MiniMax Agent is a desktop application that leverages AI to help users with creativity, productivity, meditation, podcasting, coding, and analysis. It is designed for knowledge workers seeking an intelligent assistant to streamline daily tasks.

agentflowkit is an open-source framework for constructing and managing multi-agent AI pipelines. It supports parallel DAG execution, tool calling, and cost tracking, enabling developers to efficiently build, orchestrate, and monitor complex AI workflows. Ideal for AI researchers and engineers working on agent-based systems.

mini-swe-agent is a lightweight AI agent designed for code-related tasks, emphasizing simplicity and ease of use for software engineering workflows. The tool is implemented in approximately 100 lines of Python code, with minimal dependencies, and is intended to be both understandable and hackable for research and daily development. Its architecture focuses on using only bash as its tool interface, deliberately avoiding custom tool integrations or language model tool-calling features. This approach allows mini-swe-agent to work with any language model, as all actions are executed via subprocess calls to bash, making each action independent and facilitating sandboxed execution. The agent supports a range of environments, including local setups, Docker or Podman containers, Singularity or Apptainer, Bubblewrap, and Contree, providing flexibility for deployment and experimentation. It is compatible with various language models through integrations with litellm, openrouter, and portkey, supporting endpoints such as /completion and /response, and enabling interleaved thinking. mini-swe-agent maintains a linear history, appending each step to the messages passed to the language model, which aids in debugging and fine-tuning by making the agent's trajectory transparent and easy to inspect. Designed to serve as a baseline system, mini-swe-agent is particularly suitable for researchers and developers seeking a straightforward, stable, and extensible agent for benchmarking, fine-tuning, or reinforcement learning tasks. It is presented as a convenient command line tool and also offers Python bindings for programmatic use. The tool achieves high performance on the SWE-bench benchmark and is positioned as a practical choice for those who prioritize simplicity and control over complex agent scaffolding or UI-heavy solutions. mini-swe-agent is built by the team behind SWE-bench and SWE-agent, and is intended for users who prefer a minimal, stable, and easily extensible agent to support their software engineering automation and research needs.

air-agent is an open-source Python framework for building lightweight AI agents with support for OpenAI tool calling, Model Context Protocol (MCP), and parallel subagents. It is designed for developers who want to orchestrate and extend AI agent capabilities efficiently.

fdu-micro-agent is an open-source framework for building lightweight AI agents tailored to vertical domain applications. It supports RAG, MCP, and local LLMs, enabling developers to create specialized, efficient agents for targeted use cases.