Agent frameworks & infrastructure
Agent frameworks & SDKs
12,596 agents · most starred first
Agent frameworks & SDKs, page 37 of 126
Source generator to convert C# interfaces to JSON Schema for AI tool calling
a lightweight TypeScript/Javascript library for progressively parsing streamed text (LLM/SSE-like) into structured events. Perfect for LLM powered stream-chat applications
An implementation of the OmniAI interface for Anthropic.
a small library around anthropic lib to enable claude function calling
UI frontend for loopgpt self-prompting autonomous AI agent, to aid in debugging initial prompts
The universal AI agent platform. Scaffold, configure, guard, and deploy AI agents across 7 frameworks — OpenAI, Anthropic, CrewAI, LangGraph, Google ADK, LlamaIndex, AutoGen. One command. Any model. Deploy anywhere.
HelloAGI: open AGI-agent framework with evolving character, governed autonomy, and low-latency runtime
Official plugin repository for OpenAkita — drop-in extensions for your AI team
LangChain Agent × Ollama 本地部署大模型
LLM-Compiler is an architecture that improves LLM task execution by planning, scheduling, and executing subtasks efficiently, inspired by how compilers optimize programs.
Examples to demonstrate configuration and usage of basic agents and workflows from Microsoft Agent Framework, supporting thread persistence and structured output.
🗯️LLM toolkit for RAG, tuning, agents, and more
Runtime abstractions and contracts for the UiPath Python SDK.
Innovative AI agent implementations using LangGraph—featuring ReAct, RAG (Corrective, Self, Agentic), chatbots, microagents, and more, with multi-AI agent systems on the horizon! 🤖🚀
A sophisticated multi-model AI agent framework with intelligent task routing, extensible plugin architecture, and advanced resource optimization.
A production ready AI agent framework built on finite-state machines (FSM).
Lightweight self-hosted AI agent platform. Simple / Pipeline / Handler execution modes, hybrid RAG, knowledge graph.
"From Prompt to Cognitive Engineering". — AI: Designed, not Dreamed.
TypeScript AI agent framework — 40+ LLM providers, 100+ tools, multi-agent orchestration, RAG, guardrails, circuit breakers, HITL, budget enforcement. Open-source CrewAI / LangChain alternative. — For Building Production-Grade Multi Agents in Minutes at Scale
The harness your agents grow in.
Iris Agent is a lightweight, flexible framework for building AI agents in Python. It simplifies the process of creating agents that can use tools, manage memory, and interact with various LLM providers.
Zero-code YAML-driven agent orchestration over strands-agents
All-in-One AI Playground for LLM, Chat, RAG, MCP, Agents, Skills etc.
The component framework for AI.
模块化通用 AI Agent 平台 — 持久记忆 / 自我进化 / 多智能体协作 / 知识库 / 视频生成 / IM 网关ai-agent, multi-agent, llm, knowledge-base, video-generation, im-gateway, self-evolution, persistent-memory, typescript, nextjs, fastapi, mcp, agent-framework, open-source
Agent Capability Standard is an open specification for composable AI agent capabilities. It defines 36 atomic capabilities across 9 cognitive layers, a type-safe workflow DSL, and grounded world modeling with trust-aware conflict resolution. Built on the Grounded Agency philosophy, it makes agent reliability structural—not optional.
A code-first framework for building deterministic CLI tools that LLM agents can discover, validate, and invoke with zero guesswork. Powered by Zod schemas and structured output contracts.
Apple MLX LM backend for the Agent Framework.
Eltes is an advanced AI agent harness designed for complex, agentic tasks. Features planning, filesystem sandboxing via Daytona, long-term memory, and multi-agent orchestration. Built with Vercel AI SDK, Qstash Workflow and Upstash.
⚡ Nexus Agent — Zero-config, self-evolving local AI agent framework. Multi-agent orchestration, GraphRAG, IDE integration. Runs 100% local with Ollama.
A custom reactive AI Agent framework that allows for creating flexible reactive agents to carry out tasks using MCP and Custom tools.
Full Rust SDK for the Gemini Multimodal Live API — wire protocol, agent runtime, and fluent DX in three layered crates.
Autonomous AI-agent workflow experiments for Xiaomi MiMo Orbit 100T Token Grant for Builders.
Modular open-source framework for AI agents to autonomously execute the full SDLC. 5-phase structured lifecycle with quality gates and fail-safe protocols.
Local-first .NET 10 sample integrating Microsoft Agent Framework with Foundry Local through a Microsoft.Extensions.AI IChatClient adapter.
C++20 framework for LLM application and AI agent orchestration: typed components, Chain/Graph workflows, streaming, callbacks, and unified execution.
Self-hosted enterprise AI platform for RKE2 (Kubernetes): Control Panel, tenant agents and RAG, optional API. Releases and install wiki.
gradio region:us
gradio region:us
gradio region:us
docker region:us
The Middle East's first open source agentic harness. Designed for governed self-evolution.
Core hub for LLM orchestration, AI agent development, and prompt engineering frameworks. Implementing advanced workflows, RAG systems, and autonomous agent logic for intelligent automation.
A System 2 Cognitive Middleware that replaces static data pipelines with runtime code generation. Features Gemini 3 Pro, E2B Sandboxing, and Self-Correcting Triangulation.
A deterministic control harness for AI agents. Know when to accept, retry, replan, or roll back.
Stateless CLI giving AI coding agents persistent identity, a typed lifecycle, and crash-recovery. Self-hostable, AGPL-3.0.
A self-evolving AI agent OS: Rust microkernel + Python plugin pipeline. Approval gates, evaluation gates, Docker isolation, triggers — agents that run real tasks for days without losing control. 高度可配置、自进化的智能体操作系统,支持插件化管道、多层记忆、热替换和 MCP 协议,Rust 微内核 + Python 插件管道。
A Rust SDK for Anthropic's Claude API that won't panic on you — no unwrap/expect/panic, denied at compile time. Typed structured output and tools derived from Rust structs, streaming, thinking, caching, batches.
Enterprise-grade, multi-tenant, model-agnostic, MCP-first Agent platform.
面向多租户与身份隔离的垂类Agent开发框架,包含LLM wiki+知识图谱+rag的记忆与知识系统,Jev路由决策,联网搜索,子Agent,多模态,一键对接项目接口封装工具等工具包,快速搭建面向系统的Agent服务
Production-grade agentic framework for Java. Simple, explicit, no magic.
Native Claude Agent SDK with Harness specialized become as team member.
Official TypeScript/JavaScript SDK for Cognipeer Console — OpenAI-compatible chat, batch, realtime voice, embeddings, RAG, MCP, agent tracing, and guardrails for multi-tenant AI products.
Self-hosted control plane for governed AI agents: approval-first execution, persistent context, reusable profiles and skills, and auditable results.
Unified DeepSeek Harness plugin: role-based subagent routing + per-agent evolution (prefercmd/memory as knowledge allow/deny lists), so repeated tasks start from proven commands and save tokens. Unified subagent routing and evolution: prefercmd/memory serve as knowledge allow/deny lists, saving tokens.
USB - Universal Skill Bridge: one portable skill format, installed into 16 agent runtimes with sha256-verified installers
Production-grade multi-agent AI platform with LangGraph v0.3, MCP servers, hybrid RAG (Qdrant + LlamaIndex), FastAPI WebSocket streaming, and Next.js real-time UI. Deployed on AWS EKS via Terraform.
Make any Python function crash-proof in 3 lines. The zero-daemon durability kernel for AI agents & scripts.
Industrial-grade C++23 Agent Tooling Framework GUI — provider-agnostic LLM agent with true SSE streaming (Continuo transport), tool registry, approval gate. Built on Aria (C++23 MVVM).
Python Framework for Multi-Channel AI Assistants and Agent Applications
yet another simple ai agent
Tree of Thoughts (ToT) Monte Carlo Tree Search (MCTS) reasoning engine with UCB1 exploration-exploitation balance and state backpropagation.
Linux-inspired Sequential Lock (seqlock) providing zero-contention reader loops and atomic version counter validation for writer priority.
Production-grade genpark-bpe-byte-pair-encoding-tokenizer-skill skill for AI agents
Bowyer-Watson incremental Delaunay triangulation engine constructing empty circumcircle simplex meshes over 2D spatial point clouds.
Production-grade genpark-flatbuffers-zero-copy-vtable-serializer-skill skill for AI agents
Production-grade genpark-lsm-tree-compaction-memtable-engine-skill skill for AI agents
Davis-Putnam-Logemann-Loveland (DPLL) Boolean satisfiability solver with unit propagation and recursive backtracking.
Cooley-Tukey Radix-2 Fast Fourier Transform (FFT) frequency analyzer converting discrete time-domain signals into spectral complex frequency bins.
Production-grade genpark-three-phase-commit-non-blocking-skill skill for AI agents
Production-grade genpark-flat-combining-concurrent-priority-queue-skill skill for AI agents
Consistent hashing ring with virtual node replication for balanced distributed key partition and minimum remap churn.
Loopy Belief Propagation (LBP) engine on general factor graphs with sum-product message passing and damping to solve approximate inference on cyclic topologies.
Immix mark-region garbage collector utilizing hierarchical block and line marks, fine-grained object allocation, and opportunistic mark-sweep-evacuate defragmentation.
Erlang/BEAM-inspired reduction-based preemptive actor mailbox runtime with lock-free MPSC queues, message dispatcher, and supervisor restart hierarchies.
A framework for Forward Deployed Engineers: from problem statement to a deployable AI project — every decision traced to a fact, every fact to a source.
一个用 TypeScript 构建的多智能体学习与实验场:基于 DeepSeek Responses API,从零实现强类型 Agent Loop、工具调用回填、Skills 按需加载和 SQLite 会话持久化,并以 Gexep、Lexey、Jezeh 等角色“粘合”邮件、E2B Sandbox 等能力,以及对 MCP、A2A、最小权限协作和 GraphRAG 长期记忆的探索与思考。
An open-source framework for building stateful, multi-agent AI applications. It helps developers create reliable AI workflows with memory, human-in-the-loop capabilities, tool calling, and long-running agent orchestration.
Minimal, provider-agnostic Python client for large language models, built on niquests.
JavaScript LLM agents on the XS engine, embedded in Swift. Snapshots, resident agents, confined tools.
ZpAgent - ReAct Agent (ChatBot)Project Built with LangChain 1.2 + LangGraph + FastAPI + Vue 3, this is a ReAct-paradigm intelligent agent that supports SSE streaming output, dynamic multi-tool selection, multi-turn conversational memory, and Human-in-the-Loop tool approval.
A local AI Agent workbench for branching conversations, visual context management, parallel exploration, and tool-driven project workflows.
Build on your own AI agent platform in one command. Pick the modules you want and get an app of your own — running on Chatfuel's infrastructure, with the skills your coding agent needs to keep building it.
Reactive, artifact-driven agent runtime: agents transform versioned, typed, provenance-aware artifacts inside an evolving context — no graphs to draw.
Employee — An open-source runtime for autonomous AI workers that can reason, act, learn, and operate toward long-term goals.
Make any agent harness token-native.
ReAct (Reasoning + Acting) execution engine with step-by-step observation injection and scratchpad history tracking
The provider-agnostic AI toolkit for Python
Proactive High-Stakes Intent Disambiguator & Option Clarifier. Detects under-specified user goals, assesses reversibility and risk, and constructs minimal multiple-choice clarifications before triggering autonomous execution.
Universal LLM library for Ruby. Supports OpenAI, Anthropic, Grok, Huggingface, and more.
Conversational slot filling and intent requirement state machine generating targeted questions
Quickstarts for Diagrid Catalyst, durable execution for AI agents and workflows: LangGraph, CrewAI, Strands, Pydantic AI, Mastra, MCP and more
ManusClaw: The ultimate autonomous AI agent framework. Unleash self-reasoning CLI beasts to execute code, browse the web, and dominate tasks without limits. No GUIs, pure power.
Source code for Building AI Agents with .NET — Part 1: Microsoft Agent Framework, C#, and Production-Ready Agentic Workflows
Open infrastructure for AI agents — identity, storage, and more. Self-hostable microservice suite. / 面向 AI Agent 的开放基础设施——身份、存储及更多能力。可完整自托管的微服务套件。
A statically typed framework for working with LLMs
An agent-optimization framework ✨
Production-ready AI agent runtime aligned with 12-Factor Agent principles, featuring multi-provider support (OpenAI, Gemini, Claude), advanced memory capabilities (ReasoningBank), and 100% Vercel AI SDK V5 Stream Protocol compatibility for seamless frontend integration.
a startup but not simple agent demo using google adk.
It shows how to use an agent based on LangGraph with MCP.