Agentes e multiagente

OpenAgents

openagents-org/openagents

An open-source platform providing a collaborative operating system for AI agents, enabling unified workspace management, multi-agent coordination, and network integration without vendor lock-in.

★ 4KEstrelas
⑂ 403Forks
87Problemas em aberto
TypeScriptIdioma
Apache-2.0Licença
Q@project.QualityScorePontuação editorial

Capturas de tela do projeto

Captura de tela de OpenAgents Captura de tela de OpenAgents Captura de tela de OpenAgents Captura de tela de OpenAgents Captura de tela de OpenAgents Captura de tela de OpenAgents

Visão geral

OpenAgents is a full-stack platform designed to solve the fragmentation of AI agent management. It allows users to connect various local and cloud-based coding agents into a single, persistent workspace where they can share context, files, and browser sessions. The platform consists of three core components: a browser-based Workspace for real-time human-agent collaboration, a cross-platform Launcher (CLI or Desktop App) for managing and connecting agents, and a Network SDK for developers building custom agent systems and collaboration patterns.

Principais recursos

  • Unified workspace for any connected agent
  • Multi-agent collaboration with shared context, files, and browser
  • Persistent workspace URL for bookmarking and sharing
  • Cross-platform Launcher (CLI and Desktop App)
  • Background daemon for agent lifecycle management
  • Support for 10+ coding agents including Claude Code, Codex CLI, OpenClaw, Aider, and Goose
  • Network SDK with MCP and A2A protocol support
  • Built-in tunneling to expose local dev servers

Requisitos, instalação e início rápido

CLI (macOS / Linux): curl -fsSL https://openagents.org/install.sh | bash
CLI (Windows PowerShell): irm https://openagents.org/install.ps1 | iex
Desktop App: Download directly from https://openagents.org/api/download/launcher/mac, https://openagents.org/api/download/launcher/windows, or https://openagents.org/api/download/launcher/linux-appimage.

Uso

After installing the CLI, use the 'agn' command to manage agents. Example workflow:
1. Install a runtime: agn install openclaw
2. Create an agent instance: agn create my-agent --type openclaw
3. Set credentials: agn env openclaw --set LLM_API_KEY=sk-...
4. Start the daemon: agn up
5. Connect agent to workspace: agn connect my-agent <workspace-token>

Compatibilidade do modelo e casos de uso

Compatible with any LLM backend supported by the connected agents. For example, OpenClaw supports any LLM backend; Aider supports providers like OpenAI, Anthropic, OpenRouter, Gemini, and DeepSeek; Goose supports OpenAI, Anthropic, Google, OpenRouter, and Ollama.

Observações sobre licença e riscos

Apache-2.0

Editorial verification 2026-08-09: repository URL, owner, description, license and repository statistics were reviewed. License metadata: Apache-2.0. README was fetched for the channel draft; re-check repository dependencies, releases and model terms before production use.

Lançamento e manutenção

Not stated in the repository metadata

Firecrawl

firecrawl/firecrawl

★ 161,1KTypeScript

LangChain

langchain-ai/langchain

★ 143,6KPython

RAGFlow

infiniflow/ragflow

★ 86,7KGo