Agent SDKs & frameworks
general-purpose libraries for building agents
deepseek-ai/deepseek-harness deepseek-ai/deepseek-harness is the clear leader: a plugin-first harness that makes composing and swapping every agent component easy. Before choosing, decide whether you need that maximal composability (pick Deepseek) or a specialized opinionated stack (pick one of the niche projects below).
Which one matches your setup?
Answer any of the questions — the shortlist updates as you go. Recommendations come from the capability passports below, nothing else.
Comparison matrix
Axes are extracted from each project's docs by our review pipeline; the maturity score is computed from stars, growth and commit activity — not an opinion. Click a column to sort.
| Runs in | Models | Context | Cost to run | |||||
|---|---|---|---|---|---|---|---|---|
⭐ 255.2k +6494/7d | CLIWeb app | Fixed provider | Diff only | No controls | ●●●●● | Self-hostable | ●●●●● | Free, self-hosted |
⭐ 148.1k +377/7d | CLIIDECIWeb app | BYOKOpenAIAnthropicGemini | Diff/file only | None mentioned | ●●●●● | Cloud-key usage | ●●●●● | Your API key, per-provider pricing |
⭐ 47.5k +100/7d | Web appCLI | BYOKOpenAIAnthropicGeminiOllama / localFixed provider | Diff only | None mentioned | ●●●●● | Can run fully local | ●●●●● | Your API key for cloud models; can use local models. |
⭐ 42.8k +131/7d | CLIWeb appCoding-agent plugin | BYOK | Diff only | Human approval | ●●●●● | Self-hostable | ●●●●● | Free, self-hosted |
⭐ 38.1k +149/7d | Web appCLI | BYOKOpenAIAnthropicGemini | File-level context | Human-in-the-loop | ●●●●● | Self-hostable (cloud optional) | ●●●●● | Your LLM API key (third‑party); optional CopilotKit Cloud (separate pricing). |
⭐ 33.5k +369/7d | CLIWeb app | Fixed provider | Diff + related files | Human-in-the-loop | ●●●●● | Self-hostable | ●●●●● | Requires a model provider API key (provider-dependent). |
⭐ 30.6k +125/7d | CLICoding-agent pluginWeb app | BYOKOpenAIAnthropicGeminiFixed provider | Session/tool scope | Configurable filters | ●●●●● | Cloud w/ your API key | ●●●●● | Requires Composio dashboard API key |
⭐ 30.4k +241/7d | CLIWeb app | BYOKOpenAIAnthropicGeminiOllama / local | Whole-repo analysis | Human-in-the-loop | ●●●●● | Fully local | ●●●●● | Your API key for hosted models; can run locally with self-hosted models. |
⭐ 29.9k +128/7d | CLIWeb app | BYOKOpenAIAnthropicOllama / localFixed provider | Diff only | None mentioned | ●●●●● | Fully local possible | ●●●●● | Requires provider API keys; provider billing applies. |
⭐ 28.9k +180/7d | CLIWeb appIDECoding-agent plugin | BYOKOpenAIAnthropicGemini | Repo + related files | Human-in-the-loop | ●●●●● | Self-hostable | ●●●●● | Your API keys for chosen models (per-provider billing) |
⭐ 26.8k +54/7d | CLIWeb app | BYOKOpenAIAnthropicOllama / localFixed provider | Whole-repo analysis | Filtering & ranking | ●●●●● | Local models supported | ●●●●● | Your model provider/API key, or self-hosted (local models) |
⭐ 25.8k +264/7d | Coding-agent pluginCLIWeb app | OpenAIAnthropic | Diff + related files | Incremental merge & QC | ●●●●● | Host/cloud dependent | ●●●●● | Depends on the host/provider billing (may use your host or API key) |
Capability profiles
Six axes, 0–5 each. The shape tells you the strategy: a wide hexagon is a generalist, a spike is a specialist. Showing the 8 most established — the rest are in the full catalog.
A plugin-first agent harness where every component is a plugin, enabling highly composable agent workflows.
A broad, production-ready agent framework with a large ecosystem of integrations and modular abstractions to swap models, tools, and retrievers without rewriting application code.
A lightweight, full-featured Agent Harness with a one-click Skill Hub, built-in long-term memory, knowledge graph, and self-evolution features that combine multi-channel runtime and plugin skills.
An end-to-end agent framework delivering an SDK, AgentOS runtime, and a management UI with built-in storage, observability, and human approval for running agent platforms.
Provides a multi-platform generative UI and shared-state SDK that lets the same agent run across web, mobile, and chat via the AG-UI Protocol.
A production-ready agent framework offering multi-tenancy, fine-grained permission controls, sandboxed workspaces, and an extensible middleware/event system for building trustworthy agent services.
Provides 1000+ pre-authenticated toolkits plus per-user sessions, authentication, and a sandboxed workbench so agents can discover and execute real app tools at runtime.
A batteries-included, opinionated agent harness that bundles filesystem access, sub-agents, context management, skills, and human-in-the-loop controls out of the box.
All repositories (97)
Agent skills that give coding agents taste: art direction, reference images and a browser check instead of gen…
Open-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memor…
The Frontend Stack for Agents & Generative UI. React, Angular, Mobile, Slack, and more. Makers of the AG-UI P…
Build and run agents you can see, understand and trust.
Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to…
🤗 smolagents: a barebones library for agents that think in code.
Mastra is the modern TypeScript framework for AI-powered applications and agents.
Integrate cutting-edge LLM technology quickly and easily into your apps
Open-source AI orchestration framework for building context-engineered, production-ready LLM applications. Des…
Distilly — Distill how they think into reusable Skills for any Agent or Bot. Formerly Colleague Skill（原同事 Skil…
Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.
An open-source, code-first Python toolkit for building, evaluating, and deploying sophisticated AI agents with…
How Python does AI. Agents, realtime voice, image generation, embeddings. Every model, every interface, typed…
<⚡️> SuperAGI - A dev-first open source autonomous AI agent framework. Enabling developers to build, manage &…
Agent framework and applications built upon Qwen>=3.0, featuring Function Calling, MCP, Code Interpreter, RAG,…
"OpenHarness: Open Agent Harness with a Built-in Personal Agent--Ohmo!"
Practical patterns, starters & CLI tools for loop engineering with AI coding agents. Design systems that promp…
Pocket Flow: 100-line LLM framework. Let Agents build Agents!
AI Agent Engineering Platform built on an Open Source TypeScript AI Agent Framework
Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-…
Easily fine-tune (SFT/RL), evaluate, and deploy Qwen, Gemma, or any open agentic LLM/VLM!
Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScr…
✨ All your agents and workspaces in one place, on every device you own. Track tasks on a board, accessible fro…
OpenSquilla — Token-Efficient AI Agent with same budget, higher intelligence density
dsh-routing-suite — injector + router-standard kit: install the runtime injector first, then the task-aware re…
The open-source agent harness - the runtime layer that turns an LLM into a working agent.
Agent OS: the agent gets smarter on its own. We just hold the line: Interview-gated, staged evaluation, budget…
Build distributed, production-grade, long-running agents.
The RL Bridge for LLM-based Agent Applications. Made Simple & Flexible.
The Harness of Harnesses • built for RSI: a trusted, persistent, self-evolving multi-agent ecosystem for all-d…
AG2 (formerly AutoGen): The Open-Source AgentOS.Join us at: https://discord.gg/sNGSwQME3x
Demystify AI agents by building them yourself. Local LLMs, no black boxes, real understanding of function call…
A simple yet powerful agent framework that delivers with open-source models
Agent framework for the JVM. Pronounced Em-BAY-bel /ɛmˈbeɪbəl/
ADHD — a skill for coding agents. Tree-of-thought with pruning, built on the Claude & Codex Agent SDK. Fans ou…
Code, Build and Evaluate agents - excellent Model and Skills/MCP/ACP/A2A Support
A Python library for building AI agents that leverage the full power of Google Antigravity.
A framework-agnostic, git-native standard for defining AI agents
J-Space Cognition Suite — a model-agnostic inference-time control suite for deep reasoning, long-horizon work,…
J-Space Cognition Suite — a model-agnostic inference-time control suite for deep reasoning, long-horizon work,…
verl-agent is an extension of veRL, designed for training LLM/VLM agents via RL. verl-agent is also the offici…
The Agentic Framework of the PHP ecosystem to build production-ready AI driven applications. Connect component…
A Go framework for building production agent systems with graph workflows, tools, memory, A2A, AG-UI, MCP, eva…
Official Repo for ICML 2024 paper "Executable Code Actions Elicit Better LLM Agents" by Xingyao Wang, Yangyi C…
[GenAI Application Development Framework] 🚀 Build GenAI application quick and easy 💬 Easy to interact with…
Exo is an agent + harness architecture that is fully recursive, able to safely edit all aspects of itself at r…
OctoTools: An agentic framework with extensible tools for complex reasoning
The agent framework where every run is durable, replayable, and resumable by default.
The ultimate harness for coding agents and beyond. All your agents. All your machines. One command center. Sta…
The SmythOS Runtime Environment (SRE) is an open-source, cloud-native runtime for agentic AI. Secure, modular,…
A single interface to use and evaluate different agent frameworks
Open-source AI agent harness in native Rust — GUI, CLI, headless, and webapp from one binary. Multi-provider,…
LightAgent: Lightweight Python framework for OpenAI-compatible agents with tools, memory, guardrails, tracing,…
Build your own Cowork, AI Scientist and other SoTA Agents just by editing config files. Support anthropic skil…
Build AI agents that run 100% on-device. Sub-100ms latency on Qualcomm NPU. Zero cloud dependency.
CUGA is an open-source generalist agent harness for the enterprise, supporting complex task execution on web a…
Agent communication SDK. The open-source agent communication layer for AI agents — email, WhatsApp, Slack, Dis…
CheetahClaws: A Fast and Easy-to-Use Agent Harness Infrastructure for Long-Horizon, Multi-Model, and Tool-Usin…
The llama-cpp-agent framework is a tool designed for easy interaction with Large Language Models (LLMs). Allow…
Agent harness and tookit for building AI agents and agentic applications. CLI and SDKs included
Integrate LLM in any pipeline - fit/predict pattern, JSON driven flows, and built in concurency support.
TypeScript AI SDK and agent framework with zero dependencies. Build production AI agents and multi-agent swarm…
HarnessX is a harness foundry: forge any number of agent harnesses from reusable processors and bundles, pair…
Cross-runtime skills for Claude Code, Codex, and multi-agent workflows.
ACRYL - Agent Context Relay Yielding Lifecycles. One persistent workspace, one canonical context, any coding a…
No longer maintained — active development moved to https://github.com/markpollack/agent-client
Agent framework for smart-cockpit apps · 面向智能座舱业务的 Agent 框架:零后端、权限分级、确认流、卡片编排、快慢双层、真实三方集成,加能力=加数据不加代码
Claude Code plugin: universal radial-tree exploration engine. One tree skill + swappable presets (brainstorm /…
OmniAgent (ICML 2026): the first native omni-modal agent for active video perception — a 7B agent that beats Q…
笨办法学 Agent · 亲手打造一个 harness——不用框架，32 个练习亲手写出一个 agent harness 的每一层。系列第一本，全书已完结，三本各自独立不分先后。
Open-source local AI Agent desktop app for authorized security assessment — Electron + Mastra-powered runtime,…
A nano, minimalistic, and lightweight library for building ReACT-based AI agents that use tools to solve tasks…
Abandoned, unsupported research archive for a local-first, model-agnostic LLM agent harness exploring orchestr…
CONTINUUM: Verifiable semantic recovery for long-running AI agents. Semantic checkpoints (not conversation dum…
Hosted alternatives
If running your own reviewer is more ops than you want, these managed services cover the same job.
The managed side of letta: the same stateful-agent model with long-term memory, run and persisted for you. You keep the agent design and skip hosting the server and its memory store.
Try Letta Cloud →The hosted counterpart to adk-python — build with the open-source ADK, then deploy to Google's Agent Engine runtime with scaling, sessions and tracing handled instead of running the process yourself.
Try Google Vertex AI Agent Builder →A framework-agnostic managed runtime: take agents built with LangChain, CrewAI or any SDK here and get memory, identity, tool gateways and a secure sandbox on AWS rather than assembling that infrastructure around your code.
Try Amazon Bedrock AgentCore →