kaushikb11/awesome-llm-agents

A curated list of awesome LLM agents frameworks.

What it solves

Finding the right open-source tooling for building LLM agents is overwhelming. This project collects the most relevant frameworks, runtimes, and supporting tools into a single, organized list so you can quickly compare options and pick what fits your needs.

How it works

It is a curated list (an "awesome" list) with entries organized into categories such as Core Frameworks, Multi-Agent Orchestration, CLI Agent Harnesses, Low-Code & Visual Builders, Retrieval & Data, Memory & Context, Agent Infrastructure, Safety Security & Evaluation, Domain-Specific Agents, Research & Experimental, and the 2023 wave of autonomous agents. Each entry shows the project's name, star count, primary language, license, last update date, and a one-line description. The list is generated from structured data files, and contributors can suggest new projects. Metrics refresh weekly.

Who it's for

Developers, researchers, and teams evaluating or building LLM-based agents. It is useful for anyone who wants a broad but curated overview of the open-source agent ecosystem, including frameworks in many languages (Python, TypeScript, Go, Rust, C#) and for varied use cases like coding agents, robotics, data science, and research.

Highlights

  • Covers 100+ projects across 12 categories, from core frameworks to domain-specific agents.
  • Includes both popular projects (e.g., LangChain, AutoGen, CrewAI) and niche or experimental ones.
  • Each entry provides stars, language, license, and last-update date for quick comparison.
  • Explicitly marks inactive projects so you know which are archived or unmaintained.
  • Weekly refresh keeps the data current.
  • Open to contributions via structured data files, making the list community-driven.

Related

  • Project
    docker/mcp-gateway

    A Docker CLI plugin that provides a gateway for managing and deploying isolated, containerized MCP servers for AI clients.

  • Project
    archestra-ai/OpenAPPA

    OpenAPPA is a deterministic, policy‑driven guardrail that sits between LLM agents and their tools, checking each tool call against a TOML‑based permissions policy to prevent unauthorized data flows. It offers a CLI for policy validation, integrations (e.g., Claude Code, Archestra proxy), and shows strong security benchmarks while maintaining high task‑completion rates.

  • Project
    actionbook/actionbook

    An AI browser agent that uses a Chrome extension and MCP connector to operate websites concurrently and access data behind logins to deliver finished work.

  • Project
    universal-tool-calling-protocol/code-mode

    A TypeScript library that enables AI agents to execute complex tool-calling workflows via a single block of TypeScript code instead of multiple individual tool calls.