{"data":{"slug":"zai-org-open-autoglm","name":"Open-AutoGLM","tagline":"An Open Phone Agent Model & Framework. Unlocking the AI Phone for Everyone","github_url":"https://github.com/zai-org/Open-AutoGLM","owner":"zai-org","repo":"Open-AutoGLM","owner_avatar_url":"https://avatars.githubusercontent.com/u/223098841?v=4","primary_language":"Python","stars":26040,"forks":4016,"topics":["agent","phone-use-agent"],"archived":false,"github_pushed_at":"2026-03-06T08:38:39+00:00","maintenance_label":"Slowing","stars_delta_30d":220,"url":"https://www.graphcanon.com/tools/zai-org-open-autoglm","markdown_url":"https://www.graphcanon.com/tools/zai-org-open-autoglm.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/zai-org-open-autoglm","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=zai-org-open-autoglm","description":"An Open Phone Agent Model & Framework. Unlocking the AI Phone for Everyone","homepage_url":"https://autoglm.z.ai/blog","license":"Apache-2.0","open_issues":262,"watchers":109,"ai_summary":"Open-AutoGLM is an open-source project providing a framework and model to create intelligent assistants that can automate tasks on mobile phones via natural language interaction.","readme_excerpt":"# Open-AutoGLM\n\n[Readme in English](README_en.md)\n\n
\n\n
\n

\n 👋 加入我们的 微信 社区\n

\n

\n 👋 关注智谱 AI 输入法 X 账号\n

\n

\n 🎤 进一步在我们的产品 智谱 AI 输入法 体验“用嘴发指令”\n

\n

\n AutoGLM 实战派 开发者激励活动火热进行中，跑通、二创即可瓜分数万元现金奖池！成果提交 👉 入口\n

\n\n## 懒人版快速安装\n\n你可以使用Claude Code，配置 [GLM Coding Plan](https://bigmodel.cn/glm-coding) 后，输入以下提示词，快速部署本项目。\n\n```\n访问文档，为我安装 AutoGLM\nhttps://raw.githubusercontent.com/zai-org/Open-AutoGLM/refs/heads/main/README.md\n```\n\n## 项目介绍\n\nPhone Agent 是一个基于 AutoGLM 构建的手机端智能助理框架，它能够以多模态方式理解手机屏幕内容，并通过自动化操作帮助用户完成任务。系统通过\nADB(Android Debug Bridge)来控制设备，以视觉语言模型进行屏幕感知，再结合智能规划能力生成并执行操作流程。用户只需用自然语言描述需求，如“打开小红书搜索美食”，Phone\nAgent 即可自动解析意图、理解当前界面、规划下一步动作并完成整个流程。系统还内置敏感操作确认机制，并支持在登录或验证码场景下进行人工接管。同时，它提供远程\nADB 调试能力，可通过 WiFi 或网络连接设备，实现灵活的远程控制与开发。\n\n> ⚠️\n> 本项目仅供研究和学习使用。严禁用于非法获取信息、干扰系统或任何违法活动。请仔细审阅 [使用条款](resources/privacy_policy.txt)。\n\n## 与其他自动化工具集成\n\n### Midscene.js\n\n[Midscene.js](https://midscenejs.com/zh/index.html) 是一款由视觉模型驱动的开源 UI 自动化 SDK，支持通过 JavaScript 或 Yaml 格式的流程语法，实现多平台的自动化。\n\n目前 Midscene.js 已完成对 AutoGLM 模型的适配，你可以通过 [Midscene.js 接入指南](https://midscenejs.com/zh/model-common-config.html#auto-glm) 快速体验 AutoGLM 在 iOS 和 Android 设备上的自动化效果。\n\n## 模型下载地址\n\n| Model | Download Links |\n|-------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| AutoGLM-Phone-9B | [🤗 Hugging Face](https://huggingface.co/zai-org/AutoGLM-Phone-9B)
[🤖 ModelScope](https://modelscope.cn/models/ZhipuAI/AutoGLM-Phone-9B) |\n| AutoGLM-Phone-9B-Multilingual | [🤗 Hugging Face](https://huggingface.co/zai-org/AutoGLM-Phone-9B-Multilingual)
[🤖 ModelScope](https://modelscope.cn/models/ZhipuAI/AutoGLM-Phone-9B-Multilingual) |\n\n其中，`AutoGLM-Phone-9B` 是针对中文手机应用优化的模型，而 `AutoGLM-Phone-9B-Multilingual` 支持英语场景，适用于包含英文等其他语言内容的应用。\n\n## Android 环境准备\n\n### 1. Python 环境\n\n建议使用 Python 3.10 及以上版本。\n\n### 2. 手机调试命令行工具\n\n根据你的设备类型选择相应的工具：\n\n#### 对于 Android 设备 - 使用 ADB\n\n1. 下载官方 ADB [安装包](https://developer.android.com/tools/releases/platform-tools?hl=zh-cn)，并解压到自定义路径\n2. 配置环境变量\n\n- MacOS 配置方法：在 `Terminal` 或者任何命令行工具里\n\n ```bash\n # 假设解压后的目录为 ~/Downloads/platform-tools。如果不是请自行调整命令。\n export PATH=${PATH}:~/Downloads/platform-tools\n ```\n\n- Windows 配置方法：可参考 [第三方教程](https://blog.csdn.net/x2584179909/article/details/108319973) 进行配置。\n\n#### 对于鸿蒙设备 (HarmonyOS NEXT版本以上) - 使用 HDC\n\n1. 下载 HDC 工具：\n - 从 [HarmonyOS SDK](https://developer.huawei.com/consumer/cn/download/) 下载\n2. 配置环境变量\n\n- MacOS/Linux 配置方法：\n\n ```bash\n # 假设解压后的目录为 ~/Downloads/harmonyos-sdk/toolchains。请根据实际路径调整。\n export PATH=${PATH}:~/Downloads/harmonyos-sdk/toolchains\n ```\n\n- Windows 配置方法：将 HDC 工具所在目录添加到系统 PATH 环境变量\n\n### 3. Android 7.0+ 或 HarmonyOS 设备，并启用 `开发者模式` 和 `USB 调试`\n\n1. 开发者模式启用：通常启用方法是，找到 `设置-关于手机-版本号` 然后连续快速点击 10\n 次左右，直到弹出弹窗显示“开发者模式已启用”。不同手机会有些许差别，如果找不到，可以上网搜索一下教程。\n2. USB 调试启用：启用开发者模式之后，会出现 `设置-开发者选项-USB 调试`，勾选启用\n3. 部分机型在设置开发者选项以后, 可能需要重启设备才能生效. 可以测试一下: 将手机用USB数据线连接到电脑后, `adb devices`\n 查看是否有设备信息, 如果没有说明连接失败.\n\n**请务必仔细检查相关权限**\n\n\n\n### 4. 安装 ADB Keyboard(仅 Android 设备需要，用于文本输入)\n\n**注意：鸿蒙设备使用原生输入方法，无需安装 ADB Keyboard。**\n\n如果你使用的是 Android 设备：\n\n下载 [安装包](https://github.com/senzhk/ADBKeyBoard/blob/master/ADBKeyboard.apk) 并在对应的安卓设备中进行安装。\n注意，安装完成后还需要到 `设置-输入法` 或者 `设置-键盘列表` 中启用 `ADB Keyboard` 才能生效(或","github_created_at":"2025-12-08T09:23:44+00:00","created_at":"2026-07-07T17:38:26.654907+00:00","updated_at":"2026-08-19T00:01:30.626175+00:00","categories":[{"slug":"ai-agents","name":"AI Agents","url":"https://www.graphcanon.com/categories/ai-agents","markdown_url":"https://www.graphcanon.com/categories/ai-agents.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/ai-agents"}],"tags":[{"slug":"agent","name":"agent"},{"slug":"phone-use-agent","name":"phone-use-agent"}],"trust":{"provenance":{"is_fork":false,"github_id":1112189620,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-08-19T00:01:29.700Z","maintenance":{"label":"Slowing","score":36,"methodology":"github_public_v1","releases_90d":0,"days_since_push":165,"last_release_at":null,"stars_delta_30d":220,"open_issues_delta_30d":1},"security_summary":{"status":"findings","scanner":"osv@v1","low_count":12,"high_count":0,"last_scan_at":"2026-07-11T11:12:13.300Z","medium_count":0,"scan_profile":"deps","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-08-19T00:01:30.315Z"},"languages":{"value":["python"],"source":"github.language","observed_at":"2026-08-19T00:01:30.315Z"},"license_spdx":{"value":"Apache-2.0","source":"github.license","observed_at":"2026-08-19T00:01:30.315Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":{"notes":["Requires Python 3.10 or above, ADB for Android devices, and potentially HDC for HarmonyOS NEXT versions."],"min_ram_gb":null,"requires_docker":false},"constraints":{"min_ram_gb":null,"requires_docker":false},"when_to_use":["When you need an open-source solution for developing multi-modal AI phone agents that can be controlled through natural language.","In scenarios where you are working with Android or HarmonyOS devices and want to automate tasks on these phones via a Python-based framework."],"when_not_to_use":["If your project requires support for more sophisticated user interface interactions beyond the basic automation capabilities provided by Open-AutoGLM.","When developing for non-Android or non-HarmonyOS devices, as Open-AutoGLM specifically leverages ADB and HDC tools which are primarily designed for these platforms."],"source":"enrich:decision_facts","observed_at":"2026-07-11T13:38:59.018Z"},"constraint_facets":{"min_ram_gb":null,"requires_docker":false},"decision_summary":[{"label":"Requirements","value":"Requires Python 3.10 or above, ADB for Android devices, and potentially HDC for HarmonyOS NEXT versions."},{"label":"Adopt for","value":"Open-AutoGLM is a framework and model for creating intelligent assistants capable of automating tasks on mobile phones via natural language interaction, using ADB to control the device and visual models for screen感知与操作."},{"label":"License detail","value":"The Apache-2.0 license allows for free use, distribution, and modification of the software provided that these actions maintain attribution to the original authors."}]}}