What is ai-agent-frameworks?

This repository is a large, organized reference list rather than a piece of software you install. It catalogs tools used to build AI agents, meaning systems where a language model plans steps, calls outside tools, remembers information, and works toward a goal on its own. As of its last review, the list holds 174 entries across 14 categories, both open source and commercial, and it says it is reviewed monthly to stay current. The list explains that these tools sit at different layers. An agent SDK is the toolkit for building the core agent loop, meaning the code that lets a model call tools and take actions. An orchestration or workflow framework controls how an agent moves through steps and keeps track of its state. A multi-agent framework coordinates several specialized agents working together. Beyond those, the list covers browser and computer-use agents that can operate a web browser or desktop on their own, infrastructure for the Model Context Protocol which lets agents connect to external tools, systems that give agents long-term memory, low-code visual builders, tools for making long-running agent tasks resumable, managed cloud agent services, deployment and sandbox platforms for running agent code safely, and tools for testing, evaluating, and monitoring how well an agent performs. Each entry links to its original source, such as an official code repository or product page, so readers can verify the claims themselves rather than taking the list's word for it. The project also provides the same information in machine-readable formats, a JSON file and a CSV file, for anyone who wants to process the data programmatically. Contributions are welcomed, and the project documents its own methodology for how entries are chosen, ordered, and verified. It also keeps a section for discontinued or historical tools rather than deleting them outright. The full README is longer than what was shown.