Builder tools
Frameworks, SDKs (software development kits), and code libraries for building your own agents. Everything here requires programming, which is why it is kept apart from the main leaderboard.
Week 44: Oct 26 to Nov 1, 2026.
| Rank | Change | Agent | Type | Setup | Stars this week | Stars |
|---|---|---|---|---|---|---|
| 1 | New | Open Ontologies Open Ontologies is an agent app that plans changes to a production ontology and checks their consequences for developers and data teams. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +1,361 | 2,744 |
| 2 | ↑9 | Hindsight Hindsight is an agent memory system that helps AI agents learn over time instead of only recalling chat history, for developers building agents. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +4,065 | 57.9k |
| 3 | New | RocketRide RocketRide is a pipeline engine that builds, debugs, and runs AI workflows inside a code editor for developers building AI features. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +1,920 | 19.6k |
| 4 | New | mcp-server-gateway MCP Server Gateway is a starter stack that routes AI agent tool calls through a secured gateway for enterprise teams running locally with Docker. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +368 | 563 |
| 5 | New | MasterAgent MasterAgent is a C++ framework that turns requests into validated tool actions locally, for developers embedding agents in devices. | Framework or SDK | Setup level 5 of 5: Requires programming | +455 | 967 |
| 6 | ↓3 | AutoSwarm-Shadow-Adapter AutoSwarm-Shadow-Adapter is a human review add-on that lets AI agent workflows pause for a person's sign-off without stalling. | Other | Setup level 5 of 5: Requires programming | +1,622 | 5,130 |
| 7 | ↑5 | TradingAgents TradingAgents is a multi-agent framework that simulates a trading firm of LLM analysts, researchers, and managers to evaluate stock trades for researchers and developers. | Framework or SDK | Setup level 5 of 5: Requires programming | +4,861 | 120k |
| 8 | ↓1 | dsh dsh is an open-source agent platform built on plugins, for developers who want to build and run their own AI agents. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +7,342 | 269k |
| 9 | ↓8 | harness2 A Python framework that tunes the setup around an AI agent so it does better on unfamiliar professional tasks, for developers running benchmarks. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +2,195 | 22k |
| 10 | New | RSI-Anything RSI-Anything is a research framework that lets AI agents predict outcomes, test actions, and improve over repeated rounds for developers and researchers. | Framework or SDK | Setup level 5 of 5: Requires programming | +271 | 1,101 |
| 11 | ↑5 | Moli Moli is a headless browser that lets AI agents fetch pages, extract content, and automate browser tasks with low resource use. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +1,398 | 22.1k |
| 12 | ↑12 | Scrapling Scrapling is a Python web scraping framework that collects data from websites, including JavaScript-heavy and bot-protected ones, for developers. | Framework or SDK | Setup level 5 of 5: Requires programming | +2,309 | 92k |
| 13 | New | Chaos Intelligence Chaos Intelligence is a programming language that treats AI agents as callable functions for developers building multi-step agent workflows. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +162 | 223 |
| 14 | New | TraceRoot TraceRoot is an open source platform that traces production runs of AI agents and flags failures so developers can fix them for teams building agents. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +284 | 1,284 |
| 15 | New | memvid memvid is a memory library that gives AI agents long-term recall stored in a single file, for developers who build agents. | Framework or SDK | Setup level 5 of 5: Requires programming | +748 | 17.3k |
| 16 | ↑4 | AgentScope AgentScope is a framework that helps developers build and run AI agents they can monitor and trust. | Framework or SDK | Setup level 5 of 5: Requires programming | +1,121 | 35.2k |
| 17 | New | ManimGX ManimGX is a Python animation engine that renders 3D and math videos quickly for agents and creators. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +317 | 2,449 |
| 18 | New | RL-pricing RL-pricing is a reproducibility lab that re-tests published trading AI claims under one fixed protocol for researchers and quantitative analysts. | Other | Setup level 5 of 5: Requires programming | +160 | 345 |
| 19 | New | RSIHub RSIHub is a framework that lets developers run controlled, auditable self-improvement experiments on AI agents. | Framework or SDK | Setup level 5 of 5: Requires programming | +125 | 315 |
| 20 | New | precondition-library precondition-library is a research benchmark that compares ways of choosing which saved git maintenance program to replay, for developers studying agent skill reuse. | Framework or SDK | Setup level 5 of 5: Requires programming | +132 | 311 |
| 21 | New | OpenMAIC OpenMAIC is a self-hosted platform that turns a topic or document into an interactive classroom taught by AI agents, for teachers and learners. | Agent app | Setup level 5 of 5: Requires programming | +1,005 | 42.5k |
| 22 | New | AgenticX AgenticX is a Python SDK and desktop app for building and running multi-agent teams for developers and enterprises. | Framework or SDK | Setup level 5 of 5: Requires programming | +191 | 882 |
| 23 | ↓5 | Firecrawl Firecrawl is an API that searches the web and pulls clean page content for AI agents and apps. | Framework or SDK | Setup level 5 of 5: Requires programming | +2,468 | 198k |
| 24 | New | Remory Remory is a code library that adds learned memory to an agent's compaction summary so long sessions keep going for developers. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +147 | 644 |
| 25 | New | AI Native Call Center AI Native Call Center is a self-hosted phone system that has AI voice agents answer calls and hand them to human agents when needed. | Agent app | Setup level 5 of 5: Requires programming | +121 | 517 |
| 26 | ↓1 | WeKnora WeKnora is a self-hosted knowledge platform that turns a team's documents into searchable answers, an autonomous agent, and a wiki for businesses. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +787 | 36.2k |
| 27 | New | Droidline Droidline is a tool that automates real Android phones from code, for testing apps and repeating phone tasks. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +107 | 442 |
| 28 | New | xiaoai-patch xiaoai-patch is a set of firmware patches that turns XiaoAi smart speakers into open source media players and voice assistants for DIY users. | Other | Setup level 5 of 5: Requires programming | +146 | 750 |
| 29 | ↓16 | esp32-iphone-agent ESP32 iPhone Agent is a hardware and app setup that lets an AI model operate a real iPhone for hands-free phone tasks, using a small board instead of a computer. | Agent app | Setup level 5 of 5: Requires programming | +309 | 4,219 |
| 30 | ↓8 | Shaders Shaders is a library of WebGPU visual effect components for web developers building animated backgrounds and cursor effects for websites. | Framework or SDK | Setup level 5 of 5: Requires programming | +299 | 5,779 |
| 31 | ↑6 | ReAgent ReAgent is an agent app that reconstructs C/C++ source code from compiled programs for reverse engineers and security researchers. | Agent app | Setup level 5 of 5: Requires programming | +174 | 2,407 |
| 32 | New | whatsapp-ai-agent-sdk WhatsApp AI Agent SDK is a code library that builds a personal or business AI agent on WhatsApp, running on your own machine, for developers. | Framework or SDK | Setup level 5 of 5: Requires programming | +75 | 219 |
| 33 | ↓14 | Tiny Engineer Tiny Engineer is a desk robot that shows what your AI coding agent is doing by moving, changing its face, and ringing a bell. | Other | Setup level 5 of 5: Requires programming | +131 | 1,026 |
| 34 | ↓8 | agentgateway Agentgateway is a proxy that secures, monitors, and controls traffic between AI agents, models, and tools for developers running agent systems. | Other | Setup level 5 of 5: Requires programming | +241 | 5,871 |
| 35 | New | Cairn Cairn is a general-purpose search engine that runs AI worker agents to find a path from a starting point to a goal, validated first on penetration testing. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +251 | 4,749 |
| 36 | = | xiaozhi-esp32 xiaozhi-esp32 is a firmware for a voice chatbot device that controls smart gadgets through MCP for hobbyists building AI hardware. | Agent app | Setup level 5 of 5: Requires programming | +514 | 32.5k |
| 37 | New | HarnessRouter HarnessRouter is a self-hosted service that runs several AI coding agents through one API so product teams can add agent tasks to their own apps. | Framework or SDK | Setup level 1 of 5: Works in your browser or phone | +216 | 3,863 |
| 38 | New | mathema mathema is a Python tool that checks functions against stated mathematical claims and records the results for developers who want evidence their code is correct. | Framework or SDK | Setup level 5 of 5: Requires programming | +73 | 272 |
| 39 | New | Headroom Headroom is a library, proxy, and MCP server that shrinks tool outputs, logs, and files before they reach an AI model, cutting token costs for agents. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +791 | 78.6k |
| 40 | ↓11 | OpenViking OpenViking is a context database that gives AI agents organized memory, documents, and skills they can browse like files. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +536 | 41.4k |
| 41 | ↓3 | browser-use Browser Use is an open-source Python library and cloud service that lets AI agents control a web browser to finish tasks for you. | Framework or SDK | Setup level 5 of 5: Requires programming | +688 | 120k |
| 42 | ↓1 | VoiceMem VoiceMem is a memory library that gives voice agents long-term recall of facts and emotions about the user. | Framework or SDK | Setup level 5 of 5: Requires programming | +125 | 2,825 |
| 43 | ↓4 | LangGraph LangGraph is a framework that helps developers build long-running agents that keep their state and resume after failures. | Framework or SDK | Setup level 5 of 5: Requires programming | +435 | 44.3k |
| 44 | New | EnterpriseAgentFramework ReachAI is an enterprise agent platform that lets AI query data, fill forms, and run tasks inside existing business systems like OA, ERP, and CRM. | Framework or SDK | Setup level 5 of 5: Requires programming | +65 | 1,018 |
| 45 | ↑3 | Mem0 Mem0 is a memory layer that lets AI agents and apps remember user details and past context across sessions for developers building assistants. | Framework or SDK | Setup level 5 of 5: Requires programming | +468 | 68.3k |
| 46 | New | memory-bridge MemoryBridge is a shared memory layer that lets an AI keep your context across devices and apps, so you do not repeat yourself. | Framework or SDK | Setup level 4 of 5: Guided command-line setup | +51 | 442 |
| 47 | ↓2 | OpenSandbox OpenSandbox is a sandbox runtime that lets developers run AI agents in isolated environments on their own infrastructure. | Framework or SDK | Setup level 5 of 5: Requires programming | +218 | 16.4k |
| 48 | New | Dify Dify is an open-source platform that helps teams build AI agents, workflows, and document Q&A apps and move them into production. | Framework or SDK | Setup level 5 of 5: Requires programming | +699 | 160k |
| 49 | ↓9 | crawl4ai Crawl4AI is an open-source web crawler that turns any website into clean Markdown for AI agents and data pipelines. | Framework or SDK | Setup level 5 of 5: Requires programming | +489 | 86.8k |
| 50 | ↓35 | Open Intelligent UI Open Intelligent UI is a code library that builds chat answers as interactive maps, charts, and tools for developers adding AI features to their apps. | Framework or SDK | Setup level 5 of 5: Requires programming | +72 | 907 |
No entries match these filters this week. Try a wider setup level or another task.
Also tracked
-
LangChain
LangChain is a framework that helps developers build AI agents and LLM-powered applications that connect to models and data sources.
-
OpenBB
OpenBB is an open data platform that connects financial data sources to Python, dashboards, and AI agents for analysts.
-
AutoGen
AutoGen is a framework that lets developers build multi-agent AI applications for their own domain. It is archived and no longer maintained by Microsoft.
-
crewAI
CrewAI is a Python framework that lets developers build teams of AI agents that split up and complete multi-step work.
-
Appwrite
Appwrite is an open-source backend platform that gives apps and AI agents auth, databases, storage, functions, and hosting in one place.
-
Qlib
Qlib is a Python framework that helps quant researchers build and test AI-driven investment models.
-
JeecgBoot
JeecgBoot is a low-code platform that generates business systems from a plain-language request and includes a builder for AI chat, knowledge bases, and workflows for companies.
-
CCXT
CCXT is a code library that lets developers connect to over 100 crypto exchanges and prediction markets through one interface.
-
Agno
Agno is a framework and service platform that builds, runs, and improves AI agents for developers and teams who need them available as an API or MCP server.
-
PageIndex
PageIndex is a Python library that helps an LLM find answers in long professional documents by reasoning over a document tree instead of vector search.
-
CopilotKit
CopilotKit is an open-source toolkit that puts AI agents into apps your users can chat with, click, and approve actions in, for developers building agent interfaces.
-
Cognee
Cognee is an open-source memory platform that gives AI agents persistent long-term memory across sessions for developers and teams.
-
Graphiti
Graphiti is a Python framework that builds memory graphs for AI agents, tracking how facts change over time for developers.
-
Conductor
Conductor is a workflow engine that runs multi-step jobs and AI agents reliably, so steps survive crashes and retries, for developers.
-
Composio
Composio is a developer SDK that gives AI agents pre-authenticated access to 1000+ apps for developers building agents.
-
Page Agent
Page Agent is a JavaScript library that adds an AI agent to a web page so visitors can control it with plain language.
-
Mastra
Mastra is a framework that helps developers build AI agents and AI-powered apps in TypeScript, from prototype to production.
-
FastMCP
FastMCP is a Python framework that helps developers build MCP servers and clients, connectors that let AI apps use other tools and data.
-
Haystack
Haystack is a Python framework that helps developers build production LLM apps and agent workflows for search and question answering.
-
Stagehand
Stagehand is a code library that lets AI agents use websites in a browser and pull structured data from them for developers.
-
python-sdk
python-sdk is a code library that lets developers build MCP servers and clients in Python, so AI apps can use outside tools and data.
-
NemoClaw
NemoClaw is a reference stack that runs AI agents like OpenClaw and Hermes more safely inside sandboxes for developers and technical teams.
-
Univer
Univer is an open-source SDK that lets developers embed spreadsheet, document, and presentation editing into their own products.
-
Agent Development Kit (ADK)
Agent Development Kit is a Python code library that helps developers build, test, and deploy AI agents.
-
Pydantic AI
Pydantic AI is a Python SDK that builds typed AI agents that return validated results for developers.
-
FunASR
FunASR is an open-source speech recognition toolkit that turns recorded or live audio into text for developers and teams building transcription tools.
-
notebooklm-py
notebooklm-py is an unofficial Python API and agent skill that lets developers and AI agents drive Google Gemini Notebook from code, scripts, and tools like Claude Code.
-
elizaOS
elizaOS is an open source TypeScript framework and app stack for building autonomous AI agents that run tasks across apps and devices for developers.
-
Agent Lightning
Agent Lightning is a framework that trains AI agents with reinforcement learning so they get better at their tasks for developers.
-
SkillOpt
SkillOpt is a training tool that improves reusable instruction files for AI agents without changing model weights, for developers and researchers.
-
WrenAI
WrenAI is an open-source semantic layer and agent toolkit that lets AI agents answer business questions with governed SQL and dashboards.
-
Memori
Memori is a memory layer that gives AI agents and chat apps persistent memory of past conversations and actions for developers building production systems.
-
ag-ui
AG-UI is an open protocol that connects AI agents to user-facing apps and interfaces for developers building agent-powered software.
-
Trigger.dev
Trigger.dev is a TypeScript platform that runs long AI agent tasks without timeouts for developers building agents and workflows.
-
Corsair
Corsair is a TypeScript integration platform that lets developers connect their apps and their users' apps to agents and backend services through one syntax.
-
Midscene
Midscene is a testing framework that lets developers write end-to-end tests for web, mobile, and desktop apps in plain language.
-
Agents
Agents is a framework that lets developers build realtime voice AI agents that talk, listen, and join calls for businesses.
-
E2B
E2B is an open-source sandbox service that lets developers run AI-generated code safely in isolated cloud environments.
-
Agent Framework
Agent Framework is a code library that helps developers build, connect, and deploy AI agents and multi-agent workflows in Python and .NET.
-
EverOS
EverOS is a Python library and local memory store that lets AI agents keep notes across apps and sessions, stored as readable Markdown files.
-
SeleniumBase
SeleniumBase is a Python framework that automates web browsers for testing, scraping, and stealth browsing for developers.
-
CubeSandbox
CubeSandbox is a sandbox service that runs AI agent code in isolated virtual machines for developers who need safe code execution.
-
Agent S
Agent S is a framework that lets an AI operate a computer's screen by clicking, typing, and scrolling for people who want to automate desktop and web tasks.
-
MemOS
MemOS is a memory system that gives AI agents long-term memory across sessions for developers and teams building agents.
-
camofox-browser
camofox-browser is a stealth browser server that lets AI agents load and use websites that block automated browsers.
-
CocoIndex
CocoIndex is a Python framework that keeps data for AI agents and LLM apps continuously fresh by reprocessing only what changed.
-
Spring AI Alibaba
Spring AI Alibaba is a framework that helps Java developers build agents that run multi-step tasks and coordinate with other agents.
-
VoltAgent
VoltAgent is a TypeScript framework for building AI agents with memory, tools, and multi-step workflows for developers.
-
Mobilerun
Mobilerun is an open-source framework that lets an AI agent control Android and iOS phones from natural language commands for app testing and phone tasks.
-
Strands Agents
Strands Agents is an open-source SDK for building and running AI agents in Python and TypeScript for developers who would otherwise write their own agent loop.
-
lamda
lamda is a Python client and on-device server that automates, streams, and captures traffic from Android phones for developers running device farms.
-
Honcho
Honcho is a memory library that gives AI agents persistent memory of users and projects over time for developers building stateful products.
-
Flyte
Flyte is a framework that runs ML pipelines, models, and agents reliably at scale for developers who write Python.
-
Reef
Reef is an open-source infrastructure for agents that keep improving from user feedback, for developers and researchers running model training or agent refinement.
-
OpenEvolve
OpenEvolve is an open-source evolutionary coding agent that uses LLMs to find and refine better code for a given problem for developers and researchers.
-
Swarms
Swarms is a Python framework that lets developers build teams of AI agents that split up and complete multi-step work.
-
TrueForge
TrueForge is a runtime that runs the execution loop for AI agents, with a chat UI and API, for developers and teams who want agents running reliably.
-
agentscope-java
AgentScope Java is a framework that lets developers build long-running, production-grade AI agents in Java.
-
OpenConnector
OpenConnector is a self-hosted connector gateway that lets AI agents use 1,000+ SaaS apps for developers and teams without giving agents provider credentials.
-
monid
Monid is a connector layer that gives AI agents one key to over 2,000 tools from 72 providers for web search, enrichment, and media generation.
-
Raven
Raven is a multi-agent orchestration system that coordinates specialized AI agents for complex tasks such as research, coding, and visual design for developers and teams.
-
RLinf
RLinf is an open-source reinforcement learning infrastructure for training embodied and agentic AI models at scale.
-
AG2
AG2 is a Python framework that helps developers build AI agents that work together on tasks.
-
JobSpy
JobSpy is a Python library that collects job postings from several job boards into one table for job seekers and researchers.
-
skills
skills is a collection of reusable skill files that teach AI coding agents repeatable workflows for research, content, and codebase work.
-
Google Antigravity SDK
Google Antigravity SDK is a Python library that lets developers build AI agents powered by Gemini and Antigravity without managing the agent loop.
-
TradingAgents-astock
TradingAgents-astock is a multi-agent framework that runs A-share stock research with seven AI analysts and a bull-bear debate, for researchers and developers.
-
HolmesGPT
HolmesGPT is an open-source agent that investigates production incidents and finds root causes for operations teams.
-
commerce-agents
Commerce-agents is a reference blueprint that helps developers build shopping and merchant agents for online stores with Claude.
-
HelloAgents
HelloAgents is a Python framework that gives developers reusable parts for building AI agents that call tools, search documents, and keep memory.
-
ASC
ASC is a fast, low-memory Android decompiler that lets AI agents and mobile researchers search and read app code without a long preprocessing step.
-
NitroStack
NitroStack is a TypeScript framework that helps developers build and deploy MCP servers, connectors that let AI apps use other tools.
-
Stripe AI
Stripe AI is a set of SDKs and agent skills that connect Stripe billing and payments to AI apps for developers building AI-powered products.
-
RPent
RPent is an open framework for building physical robot agents that learn from interaction with the real world for robotics developers and researchers.
-
AntOmniEvo
AntOmniEvo is a framework that automatically improves AI agent systems by having AI coding agents rewrite their tunable files based on failure results.
-
Tracely
Tracely is a self-hostable tool that turns failed AI agent runs into regression tests that block bad pull requests for developers.
-
Neuro SAN Studio
Neuro SAN Studio is a framework and playground for building multi-agent systems that split a task across several AI agents for developers and domain experts.
-
git-ai
git-ai is a version control server built for coding agents that land many small changes at once.
-
Cayu
Cayu is a Python framework that runs long-running AI agents durably, with restart recovery, approvals and spend caps, for developers building domain-specific agents.
-
C2C
C2C is a Python code library that lets two language models share their internal memory directly instead of exchanging text, for developers building multi-model systems.
-
AgentOS
AgentOS is a TypeScript framework that helps developers build AI agents that remember past conversations and create their own tools.
-
Cambium
Cambium is a governance standard and toolset that keeps LLM agents' work on a knowledge repository auditable and resumable for operators.
-
Oracle3
Oracle3 is a trading engine and MCP server that checks prediction market contracts for pricing gaps after fees, for developers and quantitative traders.
-
AppWorld
AppWorld is a benchmark and simulated environment that tests how well coding agents use app APIs for developers building and evaluating agents.
-
AgentEnv Framework
AgentEnv Framework is a Python SDK and command-line tool that builds, runs, and scores test environments for AI agents for developers building agent evaluations.
-
Cortico
Cortico is a TypeScript agent framework for building persistent persona bots, AI streamers, and companions that handle live input from several platforms.
-
OPAL
OPAL is a PyTorch optimizer wrapper that lets a coding agent rewrite training settings while a model trains, for developers training neural networks.
-
agentring
agentring is a self-hosted tool that lets your AI agent call your phone when it needs a decision, with a voice AI relaying the transcript back.
-
IDA-instances-MCP
IDA-instances-MCP is an MCP server that lets AI apps drive multiple headless IDA Pro reverse engineering sessions at once for teams hosting binary analysis.
-
pagesift
pagesift is a self-hosted tool that turns web pages, PDFs and Office files into clean Markdown for AI agents and LLMs.
-
schemagate
schemagate is a Python library that limits which database tables a text-to-SQL agent can see, based on the caller's permissions, for teams running SQL agents.
-
WOML
WOML is a markup-and-JavaScript language and runtime that builds automated workflows, backend apps, and AI agents for developers who want readable, version-controlled processes.
-
dnb-ai
dnb-ai is an Islamic-knowledge AI service that answers religious questions with cited Qur'an and Hadith references for the Deen Bridge learning platform.
-
ResolveLab
ResolveLab is a test environment that grades whether AI agents actually changed the system correctly, for researchers and developers training or evaluating tool-using agents.
-
HarborRL
HarborRL is a framework that trains tool-using AI agents with reinforcement learning on sandboxed coding and terminal tasks for developers.
-
R2D2
R2D2 is a self-hosted home screen, remote browser, and push-to-talk voice assistant for webOS TVs for people who are comfortable with root access and risky firmware modification.
-
Co-Design-Bench
Co-Design-Bench is a benchmark that tests whether coding agents correctly revise existing web applications from real user requests for developers and researchers.
-
Alter
Alter is an access control service that stores agent credentials, enforces per-action policy, and logs every call for developers.
-
ML Intern
ML Intern is a retired agent that researched and wrote machine learning code using Hugging Face tools, and is no longer maintained.
-
pi pod
pi pod is an agent app that runs your pi coding agent in isolated sandboxes on your own server for software engineers.