Agent frameworks

An MCP server that lets AI coding agents search, discover, and evaluate skills from the SkillFlow marketplace.

Hugging Face's lightweight agent framework, including code agents that write and execute Python actions.

An open-source Go client for the hosted Sonzai Mind Layer API, providing agent runtimes with memory context, turn persistence, and voice services.

A web dashboard and local runtime for orchestrating parallel AI coding agents, with task dispatch and isolated workspaces.

AWS's model-driven SDK for building agents that use tools through a model-directed loop.

Superagent is an open-source AI assistant framework backed by Y Combinator, facilitating the integration of large language models (LLM) and generative AI into applications, supporting various use cases such as question answering, chatbots, and content generation

Swarms orchestrates multi-agent collaboration for production-grade applications, solving issues like short memory and high costs, with customizable tools for specific needs, currently used by RBC, John Deere, and AI startups

This open source project introduces SWE-agent, a software engineering agent for LMs like GPT-4, enhancing bug and issue resolution in GitHub repositories with state-of-the-art performance, facilitated by a well-designed Agent-Computer Interface (ACI) and support for OpenAI and Anthropic Claude models

A self-hosted workspace for coordinating AI agent teams with structured tasks, persistent memory, and file-conflict controls.

The Forge coordinates LLMs to generate code candidates, compiles and benchmarks them, then evolves top candidates through mutation and selection.

A local-first AI agent framework for running specialist agents with shared memory, tools, approvals, and mission control.

uAgents is a Python library by Fetch AI for creating autonomous AI agents with features like easy creation, blockchain network connectivity, and cryptographic security

A self-hosted AI gateway and control plane that exposes governed MCP tools, converts OpenAPI services to MCP, and proxies compatible model APIs with virtual-key budgets, caching, and audit.

Upsonic is a Python framework for building autonomous AI agents, with tools for workflows, MCP, and agent evaluation.

Vetto is a daemon-less, rootless sandbox that applies kernel-enforced security policies to AI coding agents, with agent command shims and MCP tool isolation.

An MCP-native task control plane for AI agent teams, with persistent tasks, scoped permissions, delegation, and provenance.

The Zapier SDK is a TypeScript SDK and CLI for building agent tools and integrations against Zapier’s connection layer, giving coding agents programmatic access to Zapier’s app catalog and action library.