OpenAgent开源AI工作流 是 AI Skill Hub 本期精选AI工具之一。综合评分 7.2 分,整体质量较高。我们推荐使用将其纳入你的 AI 工具库,帮助提升工作效率。
自托管AI智能体操作系统,集成流式对话、工具调用、持久化记忆和多智能体协作功能。适合需要构建复杂AI工作流、本地部署LLM应用的开发者和企业用户。
OpenAgent开源AI工作流 是一款基于 Python 开发的开源工具,专注于 多智能体、自托管、AI工作流 等核心功能。作为 GitHub 开源项目,它拥有活跃的社区支持和持续的版本迭代,代码完全透明可审计,支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流,都能提供稳定可靠的解决方案。
自托管AI智能体操作系统,集成流式对话、工具调用、持久化记忆和多智能体协作功能。适合需要构建复杂AI工作流、本地部署LLM应用的开发者和企业用户。
OpenAgent开源AI工作流 是一款基于 Python 开发的开源工具,专注于 多智能体、自托管、AI工作流 等核心功能。作为 GitHub 开源项目,它拥有活跃的社区支持和持续的版本迭代,代码完全透明可审计,支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流,都能提供稳定可靠的解决方案。
# 方式一:pip 安装(推荐)
pip install openagentd
# 方式二:虚拟环境安装(推荐生产环境)
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install openagentd
# 方式三:从源码安装(获取最新功能)
git clone https://github.com/lthoangg/OpenAgentd
cd OpenAgentd
pip install -e .
# 验证安装
python -c "import openagentd; print('安装成功')"
# 命令行使用
openagentd --help
# 基本用法
openagentd input_file -o output_file
# Python 代码中调用
import openagentd
# 示例
result = openagentd.process("input")
print(result)
# openagentd 配置文件示例(config.yml) app: name: "openagentd" debug: false log_level: "INFO" # 运行时指定配置文件 openagentd --config config.yml # 或通过环境变量配置 export OPENAGENTD_API_KEY="your-key" export OPENAGENTD_OUTPUT_DIR="./output"
<p align="center"> <img src="https://raw.githubusercontent.com/lthoangg/openagentd/main/documents/assets/brand/openagentd-primary-lockup.png" alt="OpenAgentd octobot mascot and wordmark" width="760"> </p>
Your on-machine multi-agent system. A long-running local service with a web cockpit, persistent memory, and a team of agents that coordinate to get real work done. Everything stays on your hardware.

---
| Section | Contents |
|---|---|
| [Install](https://github.com/lthoangg/openagentd/blob/main/documents/docs/install.md) | pip, uv, Homebrew, Docker, source |
| [Migration](https://github.com/lthoangg/openagentd/blob/main/MIGRATION.md) | Move setup from OpenClaw, Hermes, Claude Code, Codex CLI, or older OpenAgentd installs |
| [CLI reference](https://github.com/lthoangg/openagentd/blob/main/documents/docs/cli.md) | Every openagentd subcommand |
| [Configuration overview](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration.md) | Hub — links into the focused subpages below |
| [Environment variables](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/env.md) | Settings fields, provider keys, optional extras |
| [Paths & XDG roots](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/paths.md) | DATA / CONFIG / STATE / CACHE / WORKSPACE / WIKI |
| [LLM providers](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/providers.md) | Every registered prefix, OAuth flows, capability YAML |
| [Agent files](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/agents.md) | .md frontmatter schema, validation, editing workflow |
| [Built-in tools](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/tools.md) | Filesystem, shell, web, multimodal, memory |
| [Skills](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/skills.md) | SKILL.md format, builtin skill catalog |
| [Sandbox & permissions](https://github.com/lthoangg/openagentd/blob/main/documents/docs/configuration/sandbox.md) | Denylist paths, user sandbox.yaml, permission services |
| [Comparison](https://github.com/lthoangg/openagentd/blob/main/documents/docs/comparison.md) | How OpenAgentd compares to opencode, openclaw, hermes-agent |
| [Troubleshooting](https://github.com/lthoangg/openagentd/blob/main/documents/docs/troubleshooting.md) | Common install and runtime issues |
Desktop app (double-click install, no terminal) — download from the latest release:
| Platform | Artefact | First-launch note |
|---|---|---|
| macOS (Apple Silicon) | brew install --cask lthoangg/tap/openagentd *or* OpenAgentd_*_aarch64.dmg | The cask ad-hoc signs and installs automatically. With the .dmg, run the bundled install.sh then right-click → **Open**. The app is unsigned — [why](https://github.com/lthoangg/openagentd/blob/main/documents/docs/install.md#desktop-unsigned). |
| Windows | OpenAgentd_*_x64-setup.exe | SmartScreen warns on first run — click **More info → Run anyway**. |
| Linux | OpenAgentd_*_amd64.AppImage | chmod +x the AppImage, or run the bundled install.sh --install for a launcher entry. |
macOS — after mounting the .dmg:
cd /Volumes/OpenAgentd*
./install.sh --install # ad-hoc signs + copies to /Applications
Then right-click OpenAgentd.app → Open the first time (single-click won't work).
CLI / API server (terminal install — ships the same UI on http://localhost:4082):
```bash
| Home | Coding workspace |
|---|---|
|  |  |
| Agents settings | Sandbox settings |
|---|---|
|  |  |
| Voice settings | Telemetry dashboard |
|---|---|
|  |  |
---
功能完整的开源AI智能体框架,具备持久化记忆和多智能体特性,但社区规模有限,选型需谨慎评估长期维护。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议,可商用,需保留版权声明和 NOTICE 文件,含专利授权条款。
经综合评估,OpenAgent开源AI工作流 在AI工具赛道中表现稳健,质量良好。如果你已有明确的使用需求,可以直接上手体验;如果还在评估阶段,建议对比同类工具后再做决策。
| 原始名称 | OpenAgentd |
| 原始描述 | 开源AI工作流:Self-hosted AI agent OS — streaming chat, tool use, persistent memory, and multi。⭐174 · Python |
| Topics | 多智能体自托管AI工作流持久化记忆工具调用 |
| GitHub | https://github.com/lthoangg/OpenAgentd |
| License | Apache-2.0 |
| 语言 | Python |
收录时间:2026-05-21 · 更新时间:2026-05-22 · License:Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。