AI Agents

Build ChatGPT over your data, all with natural language

Multi-agent collaboration system built on large language models, letting users drive complex tasks from intent to delivery through everyday chat applications.

Autonomous agent designed to sustain itself by securing compute resources, improving its capabilities, and replicating, while continuously performing real-world tasks without human intervention.

Local-first AI agent workspace for developers, writers, designers, and researchers, unifying coding, writing, design, research, and automation tasks in one runtime with desktop GUI and TUI.

WhatsApp MCP server

Desktop screen-aware AI assistant triggered by hotkey, combining current application context with multiple models and MCP tools to help users perform cross-application tasks without switching windows.

An open-source desktop-grade office agent from Youdao that operates the real desktop for data analysis, documents, slides, video tasks, and remote messaging commands.

Combines an open-source AI-native vector design tool with prompt-to-UI generation on a live canvas, supporting design-as-code output and concurrent multi-agent teams as an alternative to Pencil.

Delivers a self-hostable personal AI assistant combining multiple models, RAG, and agent loops for computer use, browsing, coding, visual workflows, and MCP tool calls.

A local-first AI desktop assistant and workspace that records model calls and task execution in a unified logging runtime, helping individual users run and audit personal automation workflows privately.

A high-performance agent workspace where desktop, TUI, and CLI clients run autonomous tasks through a unified runtime and event log.

Multi-agent personal assistant organized around org-chart teams, sharing long-term memory and collaborating autonomously while keeping secrets local for task execution.

Fork of a general autonomous agent that avoids API keys by driving major models through logged-in browsers, executing local tool tasks to reduce costs.

Combines a generalist AI agent, workflow engine, messaging, and collaborative office tools into an open-source platform for enterprises automating around-the-clock operations.

Builds runnable integration code by reverse-engineering platforms internal APIs, enabling developers to create permissionless third-party integrations without official APIs or manual scraping.

Targets professional development teams with an enterprise AI platform that combines cloud development environments, model management, task tracking, requirements handling, and autonomous coding assistance.

On-device privacy-focused mobile AI agent integrating mainstream models with a sandboxed browser and device controls to autonomously complete everyday work and life tasks.

Agentic LLM for autonomous data science, handling specific data tasks and data-oriented deep research while producing analyst-grade reports for researchers and data teams.

Self-hosted multi-user, multi-agent AI assistant designed for parallel collaborative office work, with plugins, connectors, memory, knowledge bases, and local privacy-preserving deployment.

The easiest way to use Agentic RAG in any enterprise

A personal terminal-based AI agent for coding and knowledge work, with shell, code, and web tools plus persistent autonomous runs.

Creates an open-source multi-agent platform where each agent receives identity, memory, and workspace, enabling teams to collaborate like digital coworkers.

Open-source, vision-first browser agent

Provides a persistent local and remote agent environment for software development that retains context across sessions, self-improves, and supports unattended scheduled tasks.

Memory-first, self-improving multi-agent orchestration harness that coordinates specialized research, coding, design, and monitoring agents to deliver complex tasks for teams and builders.

Assists developers by learning their coding preferences to deliver full-stack features, fix bugs, and refactor code, positioning itself as a coding agent optimized for open models.

A task-oriented AI agent productivity platform that isolates files and memory by workspace across multiple projects while continuously advancing assigned tasks.

Agentic IDE capable of controlling and customizing its own automation, driving threaded tasks across desktop, web, and CLI while delegating to other agents for developers.

Self-building agent IDE unifying desktop, web, and CLI interfaces to dispatch threaded workers that execute software-development tasks autonomously for developers.

Localized Chinese version of the open-source OpenClaw personal AI assistant, supporting multiple LLMs and chat platforms such as WhatsApp, Telegram, and Discord for automated task execution.

Multi-agent crew for managing a second brain that automatically organizes, archives, and retrieves notes, email, and transcriptions across knowledge, nutrition, and wellness tasks.

Streamer-Sales 销冠 -- 卖货主播 LLM 大模型🛒🎁，一个能够根据给定的商品特点从激发用户购买意愿角度出发进行商品解说的卖货主播大模型。🚀⭐内含详细的数据生成流程❗ 📦另外还集成了 LMDeploy 加速推理🚀、RAG检索增强生成 📚、TTS文字转语音🔊、数字人生成 🦸、 Agent 使用网络查询实时信息🌐、ASR 语音转文字🎙️、Vue 生态搭建前端🍍、FastAPI 搭建后端🗝️、Docker-compose 打包部署🐋

Open-source AI workbench for scientific research that reads literature, writes and runs code, conducts experiments, and logs the full process across desktop, browser, and terminal.

Rebuilds the OpenClaw agent gateway in Go as a single binary, supporting multi-tenant isolation, multiple LLMs and messaging channels, and orchestrated agent teams for scalable deployments.

Multi-agent investment-research framework adapted for China's A-shares, where seven analyst agents debate under A-share rules using local sources such as block-trade and unlock data.

Creates a conversational agent that learns from dialogue and evolves over time, using a memory layer and skill evolution to continuously improve behavior.

Delivers a cross-platform open-source voice assistant framework with real-time speech interaction, multimodal vision tasks, MCP integrations, and IoT device control for desktops and embedded devices.

AI-agent penetration-testing framework for security engineers that orchestrates information gathering, vulnerability discovery, exploitation, and report generation from natural-language input using LLMs and MCP tooling.

Supplies an open-source framework for building and deploying autonomous agents that handle website creation, mobile app generation, deep research, and multi-application workflows.

Local-first, open-source AI content discovery agent that learns user preferences and proactively surfaces relevant posts and videos from Bilibili, Xiaohongshu, Douyin, YouTube, X, Zhihu, Reddit, Weibo, and the open web.

Presents a self-evolving AI engineering system where a cross-model agent swarm lives inside applications, handling ideation, building, review, and self-healing through persistent neural-link integration.

Personal investment copilot that monitors portfolios against full-market data, produces research insights, and reasons about dynamic rebalancing to support individual investors making allocation decisions.

A persistent AI assistant designed for 24/7 operation via CLI and Telegram, featuring a configurable persona, permission-hardened tools, long-term memory, and token budgets for controlled autonomous work.

General-purpose AI state-space search engine using blackboard architecture to schedule agents for autonomous exploration. First validated in autonomous penetration testing scenarios.

An open-source platform for persistent AI teammates with memory, voice, and sandboxed computer access that can delegate work to peers.

A secure persistent personal agent server distributed as a single Rust binary, supporting multiple LLM providers, sandboxed execution, memory, voice, and integrations with Telegram, WhatsApp, Discord, Teams, and MCP tools.

Open-source offensive-security harness that connects to any LLM and autonomously performs reconnaissance, vulnerability discovery, exploitation, and structured penetration-test reporting workflows for security teams.

Implements an agentic vulnerability management harness covering threat modeling, detection, validation, and remediation, producing structured security findings for engineering teams.

Autonomous AI red-team framework automating offensive security operations from reconnaissance and exploitation to post-exploitation review, generating remediation reports and fix pull requests without human intervention.

A local-first command-line and terminal-interface coding agent that runs open-weight models entirely on the user's machine through a llama.cpp fork, supporting private offline software development without cloud inference.

Local-first terminal coding agent built with React and Ink, supporting multiple models and autonomous development modes to read, write, and modify code directly for developers.

A local-first personal AI desktop system with persistent memory and background processing. It executes online tasks for users and participates in a peer-to-peer marketplace for agent skills.

An AI full-stack engineer agent workspace built with Next.js and PostgreSQL using Kubernetes infrastructure. It offers persistent environments, memory files, and isolated execution for engineering tasks.

Open-source, self-hostable AI-powered security operations center providing alert fusion, purple-team drills, agent-assisted triage, and MITRE ATT&CK-based investigation under MIT license.

An MCP server for interacting with the Financial Datasets stock market API.

Builds a personal AI agent with persistent memory that manages conversations, tasks, memory, and approvals across devices, supporting continuous learning and coordinated execution for individual users.

Web Crawling and RAG Capabilities for AI Agents and AI Coding Assistants

Terminal AI coding assistant optimized for deepseek-v4, supporting file reading, editing, execution, adjustable reasoning depth, and Agent Skills extensions. Targets developers working in command-line workflows.

A natural-language data assistant built with LangChain and LangGraph using multi-agent collaboration and MCP skills. It converts user questions into queries and end-to-end data insights.

General-purpose AI agent assistant for professionals supporting Claude and Pi runtimes, multiple LLM providers, and Feishu group-chat collaboration integrated into everyday workflows.

A local-first AI agent desktop client that operates on files, executes commands, manages scheduled tasks, and supports Web UI access plus remote gateway expansion.

Security platform orchestrating multiple AI agents to discover, validate, rank, and report real code vulnerabilities with exportable findings for development teams.

Desktop coding-agent hub for GitHub repositories that discovers tasks, runs parallel agent sessions, steers implementation, and handles review and merging.

Local-first multi-agent desktop application uses a commander to coordinate specialist research, coding, data-analysis, document, and media agents, supporting bring-your-own model keys under MIT licensing.

Runs an autonomous coding agent entirely locally around small language models, helping developers plan, validate, and repair code without relying on hosted APIs or usage billing.

An open-source desktop AI agent workspace for research, design, and coding that uses tools, files, and knowledge to deliver finished artifacts.

Command-line tool for fully automatic long-form novel generation using multiple cooperating agents handling outlining, drafting, proofreading, and arbitration for coherent output.

Command-line agent client for MCP servers integrating LLM chat, tool invocation, and autonomous planning execution. Designed for developers testing and operating MCP-based agent workflows.

Automates end-to-end product growth by analyzing code and product context, gathering data, researching markets, executing campaigns, and reviewing real results to improve subsequent iterations.

Always-on personal AI operating system centered on memory and knowledge graphs for task planning. Drives coding sessions and task execution toward delivered outcomes.

Self-contained Rust agent harness living in the shell, offering Claude Code-like coding assistance in a single optimized binary without heavy dependencies or model restrictions.

Self-contained terminal agent harness distributed as a single Rust binary with zero dependencies, working like Claude Code or Codex across arbitrary models for efficient coding.

Runs as a personal autonomous desktop agent across Windows, macOS, Linux, and mobile systems, pursuing user-defined goals through multi-step desktop and browser automation, teams, and recurring tasks.

Fully private voice assistant running locally on the user computer offline, offering natural conversation, persistent memory, location awareness, web checks, Chrome control and extensible MCP tool support.

A local-first personal AI assistant you run and own, with a tiny readable loop, SQLite memory, live dashboard, and built-in evals across multiple model providers.

Open-source autonomous DevOps agent for securing, deploying, and maintaining production-ready infrastructure. Resides on machines to monitor applications and provide alerts when needed.

Minimal ~1000-line Python AI coding agent inspired by Claude Code, supporting file operations, command execution, and sub-agent spawning. Works with any LLM for learning and forking.

System for building a personal AI-native company, automatically forming AI teams, assigning tasks, and learning continuously to complete end-to-end work.

A local-first, model-agnostic AI research workbench for desktop that unifies agents, notes, literature, charts, and reproducible workflows across macOS, Windows, and Linux.

Local terminal coding agent that reads and writes repositories, executes commands, and lets users choose their own models and rules for autonomous development tasks.

AI-powered OSINT investigation agent with twenty reconnaissance tools, an interactive REPL, CLI, and MCP server. Works with Claude, GPT-4, or local models to assist authorized security researchers with structured intelligence gathering workflows.

An agentic research platform for quantitative analysts that mines, evolves, and validates alpha factors from natural-language directions using LLM intelligence combined with evolutionary strategies.

Self-hosted, LLM-agnostic agent platform connecting enterprise systems across global and China environments through one agent core. Provides chat, planning, and centralized orchestration portal.

Python-based harness for AI coding workflows, offering multi-model routing across GPT, Claude, DeepSeek, Qwen, Kimi, GLM and MiniMax, plus stateful sessions, tool governance, and traceable delivery.

On-device autonomous agent for Chinese Android systems such as ColorOS and HyperOS, supporting cross-application GUI operations plus direct system tools, root shell, and Linux terminal capabilities.

Local-first personal AI assistant orchestrating subagents, tools, memory, voice, vision, shell, browser automation, scheduled tasks, health tracking, and messaging while keeping data on-device with optional cloud models.

Agent-driven vulnerability discovery platform for security researchers, automating source code auditing, flaw verification, and generation of CVE-style reports in one workflow.

A terminal-based agentic assistant for offensive security with human-in-the-loop control. Automates reconnaissance, enumeration, and validation to help pentesters produce evidence-backed vulnerability reports.

Personal assistant agent running inside iMessage with persistent memory, sub-agents, automations, and third-party integrations, letting users choose between Claude Agent SDK and Codex app-server runtimes.

Personal AI assistant for desktop, mobile, and web that controls computers using voice and vision, combining persistent memory with an interactive holographic persona.

Cute open-source AI desktop robot based on M5Stack CoreS3, combining conversational agents, expressive servo-driven movements, and smartphone remote control for makers.

Resident autonomous research agent that continuously modifies code, trains models, and validates deep learning experiments using Leader-Worker architecture. Maintains constant-size memory for 24/7 operation.

Open-source personal learning agent with layered memory and proactive scheduling, available across desktop, web, and terminal interfaces for continuous study assistance.

Browser-based desktop operating environment controlled through natural language, where AI agents operate applications directly. Extensible for building automated app workflows.

A local-first command-line coding agent and meta-prompting system supporting planning, implementation, verification, and worktree-isolated development for long reliable autonomous runs.

Delivers a Rust-based open-source agent harness that runs GUI, CLI, headless, and web interfaces from one binary, supporting multiple providers, MCP, skills, plugins, and agent teams.

Implements an autonomous work-seeking agent that accepts assignments, completes deliverables, handles payment, and iteratively improves its performance on paid tasks over time.

An open-source general-purpose AI agent optimized for low token consumption, supporting multiple models, self-evolving skills, and integrations with popular messaging platforms for everyday automation tasks.

Self-hosted AI on-call triage starter for Slack and Microsoft Teams operators, built with AG-UI and LangGraph to triage incidents, handle approvals, assist research, and support forkable customization.

Autonomous penetration-testing system that orchestrates agents across reconnaissance, exploitation, and reporting. Built for security practitioners, it automates end-to-end offensive workflows in a specialized agent stack.