AwwO 本地版 0.1.4 — AwwO Local for Windows

AwwO 本地版 0.1.4 — AwwO Local for Windows. Run AwwO on your own Windows computer: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits — or, if you like, with your online AwwO account's models.

在自己的 Windows 电脑上运行 AwwO：AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度；也可以接入你线上 AwwO 账号的模型。

New in 0.1.4 / 0.1.4 更新

• Seven built-in design skills, adapted from open-design and refero_skill, for nodes that deliver a page or a design: design craft, stripping the AI look, visual directions, UI states and the accessibility floor, motion discipline, a five-lens design jury, and procedural three.js scenes. Add them to a node under 技能 (Skills), like any skill. Page nodes may also load three.js r186 as ES modules.
  七个内置设计技能（改编自 open-design 与 refero_skill），供交付网页或设计的节点使用：设计工艺、去掉 AI 味、视觉方向库、界面状态与无障碍底线、动效纪律、设计评审团、程序化三维场景。和其他技能一样，在节点的「技能」里添加。页面节点还可以用 ES 模块加载 three.js r186。
• A node that names the design-craft or anti-ai-slop skill, as the pages the online AwwO's canvas assistant plans now do, runs here too (0.1.3 refused it as a skill it did not have). Where those two would not fit the model's context, the run goes without them instead of being refused.
  带有「设计工艺」或「去掉 AI 味」技能的节点（线上 AwwO 的画布助手规划的页面现在会带上这两个技能）在本地版也能运行（0.1.3 会因为没有这两个技能而拒绝）。这两个技能放不进模型上下文时，运行会去掉它们，而不是被拒绝。
• Those two skills accept the note they ask for in the hand-over as an HTML comment, so a page delivered as a single HTML file can carry it.
  这两个技能要求在交付说明里写的那句话，现在也可以写成 HTML 注释，单个 HTML 文件交付的页面也能带上。
• An edit's sound can be made by a page (its window.renderAudio), on the same frame clock as a film made as code.
  剪辑里的声音可以由页面生成（页面的 window.renderAudio），与用代码制作的影片使用同一套帧时钟。
• If you have 0.1.1, 0.1.2 or 0.1.3, install 0.1.4 over it: your data is kept.
  装过 0.1.1、0.1.2 或 0.1.3 的话，直接安装 0.1.4 覆盖即可，数据会保留。

Install / 安装

1. Download AwwO-Local_x64-setup.exe and run it. Windows 10 and 11, x64. It installs for the current account only and needs no administrator rights. It takes about 500 MB on disk, plus your data.
   下载 AwwO-Local_x64-setup.exe 并安装。适用于 Windows 10 和 11（x64）；只装给当前账号，不需要管理员权限；约占 500 MB 磁盘（另加你的数据）。
2. Open AwwO Local. Its first start sets up the database on this computer and opens admin-credentials.txt, which holds the first account's email and password.
   打开 AwwO Local。第一次启动会在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。
3. Sign in. Then add your own model API keys under 「我的引擎」 (My engines), or an AwwO 云端 connection key: make one on awwo.clawhunt.store under 设置 → 本机版连接.
   登录后在「我的引擎」填写自己的模型 API Key，或填入 AwwO 云端的连接密钥（在 awwo.clawhunt.store 的「设置 → 本机版连接」生成）。

Calls made through AwwO 云端 run on the online AwwO's models that the workspace may use (Amazon Bedrock) and count as that workspace's calls. The online AwwO records each call's usage (the model, token counts, timings, an estimated cost and any error code), not its prompts or answers; see clause 08 of the Privacy Policy. A key lasts 90 days, and you can revoke it on the same page at any time.
通过 AwwO 云端的调用在线上 AwwO 运行，使用该工作区可用的模型（Amazon Bedrock），计入该工作区。线上 AwwO 只记录每次调用的用量（模型、token 数、耗时、预估费用和错误码），不保存提示词和回答，详见隐私政策第 08 条。密钥有效 90 天，可以随时在同一页面撤销。

Your data is in %LOCALAPPDATA%\store.clawhunt.awwo.local.

• Updates and uninstalls keep it. The uninstaller removes it only if you ask it to, and asks again.
• Your canvases, runs and files are stored there.
• What a node sends to a model or media service goes to the provider whose key you added, or, through an AwwO 云端 connection, to the online AwwO.

数据保存在 %LOCALAPPDATA%\store.clawhunt.awwo.local。

• 升级和卸载都不会删除；卸载时可以选择删除，会再确认一次。
• 画布、运行记录和文件保存在这里。
• 节点发给模型或媒体服务的内容会发到你填入 Key 的服务商；通过 AwwO 云端连接时，发到线上 AwwO。

Install it inside your own user folder or %LOCALAPPDATA% (the default is), in a path of at most 80 characters without % or !. A Chinese user name or folder is fine. The AwwO menu restarts it, opens it in your browser, opens the data folder and the logs, and resets the sign-in password.
安装位置要在自己的用户目录或 %LOCALAPPDATA% 之内（默认位置就是），路径不超过 80 个字符，不含 % 或 !；用户名或路径里有中文也可以。菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg (and Chrome or Edge) installed. / 渲染成视频和剪辑节点需要自行安装 ffmpeg（以及 Chrome 或 Edge）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。

For a Mac with Apple silicon, AwwO Local is a separate download: AwwO for Mac 0.10.1. It runs the same AwwO Local as this release, built from the same commit.
Mac（Apple 芯片）版是单独的下载：AwwO for Mac 0.10.1。它运行的本地版与本次发布相同，由同一个提交构建。

Notes / 说明

The installer is not signed yet, so Windows may warn about an unknown publisher. It carries:

• every third-party licence (licenses/, build/pgsql/LIBRARIES.md), with open-design's and refero_skill's for the design skills;
• the source of the LGPL libraries libintl and libiconv (build/pgsql/sources/);
• beside PostgreSQL, Microsoft's redistributable Visual C++ runtime (14.44), from Visual Studio's redistributable folder.

安装包未签名，Windows 可能提示未知发布者。安装包内附带：

• 全部第三方许可证（licenses/、build/pgsql/LIBRARIES.md），其中包括设计技能所改编的 open-design 与 refero_skill 的许可证；
• LGPL 库 libintl 与 libiconv 的源码（build/pgsql/sources/）；
• PostgreSQL 旁的微软可再分发 Visual C++ 运行库（14.44），取自 Visual Studio 的可再分发文件夹。

SHA-256 of AwwO-Local_x64-setup.exe: df8d2974e47b3f7b235cd98f11fed2acb8b895933d4ca16ffe8e28fa9e7ad5a7 (also in SHA256SUMS.txt).
Built from AwwO commit b60b7dec3638daddf66b7074676e888e4299cae3.

AwwO 本地版 0.1.3 — AwwO Local for Windows

AwwO 本地版 0.1.3 — AwwO Local for Windows. Run AwwO on your own Windows computer: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits — or, if you like, with your online AwwO account's models.

在自己的 Windows 电脑上运行 AwwO：AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度；也可以接入你线上 AwwO 账号的模型。

New in 0.1.3 / 0.1.3 更新

• The Academy's 从这里开始 (Start here) has six how-tos, numbered the same way on their title cards, 1 / 6 to 6 / 6. Two are new: inside a node (its tasks, 让助手修改, running it again), and the settings centre.
  学院「从这里开始」现在有 6 个使用教程，片头统一编号为 1 / 6 到 6 / 6。新增两个：节点内部（任务、让助手修改、再次运行）和设置中心。
• Running the whole canvas asks first when nodes already have their result: run only the nodes without one (and what they feed), run every node again, or cancel.
  运行整张画布时，如果有节点已经有结果，会先询问：只运行还没有结果的节点（以及它们供给的节点）、全部重新运行，或者取消。
• A page that keeps data in localStorage, sessionStorage or cookies now works in the preview, which gives it an in-memory stand-in, instead of failing there and being repaired again and again.
  用 localStorage、sessionStorage 或 cookie 保存数据的页面现在可以在预览里正常运行（预览为它们提供内存中的替代），不再在预览中报错并被反复自动修复。
• Ask the canvas assistant to make one node depend on another and it wires them, instead of asking which field to use.
  让画布助手把一个节点设为依赖另一个节点时，它会直接连好，不再反问用哪个字段。
• Pressing 编辑 in a template's guide no longer leaves Backspace deleting the node.
  在模板说明里点「编辑」之后，按退格键不再误删节点。
• When a model provider refuses a request, it is called too long for the model only when the provider's own error says so; a 422 answer counts as an ordinary model error.
  模型服务商拒绝请求时，只有服务商自己的错误信息表明超长，才判定为超出模型上下文；422 回复按普通模型错误处理。
• If you have 0.1.1 or 0.1.2, install 0.1.3 over it: your data is kept.
  装过 0.1.1 或 0.1.2 的话，直接安装 0.1.3 覆盖即可，数据会保留。

Install / 安装

1. Download AwwO-Local_x64-setup.exe and run it. Windows 10 and 11, x64. It installs for the current account only and needs no administrator rights. It takes about 500 MB on disk, plus your data.
   下载 AwwO-Local_x64-setup.exe 并安装。适用于 Windows 10 和 11（x64）；只装给当前账号，不需要管理员权限；约占 500 MB 磁盘（另加你的数据）。
2. Open AwwO Local. Its first start sets up the database on this computer and opens admin-credentials.txt, which holds the first account's email and password.
   打开 AwwO Local。第一次启动会在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。
3. Sign in. Then add your own model API keys under 「我的引擎」 (My engines), or an AwwO 云端 connection key: make one on awwo.clawhunt.store under 设置 → 本机版连接.
   登录后在「我的引擎」填写自己的模型 API Key，或填入 AwwO 云端的连接密钥（在 awwo.clawhunt.store 的「设置 → 本机版连接」生成）。

Calls made through AwwO 云端 run on the online AwwO's models that the workspace may use (Amazon Bedrock) and count as that workspace's calls. The online AwwO records each call's usage (the model, token counts, timings, an estimated cost and any error code), not its prompts or answers; see clause 08 of the Privacy Policy. A key lasts 90 days, and you can revoke it on the same page at any time.
通过 AwwO 云端的调用在线上 AwwO 运行，使用该工作区可用的模型（Amazon Bedrock），计入该工作区。线上 AwwO 只记录每次调用的用量（模型、token 数、耗时、预估费用和错误码），不保存提示词和回答，详见隐私政策第 08 条。密钥有效 90 天，可以随时在同一页面撤销。

Your data is in %LOCALAPPDATA%\store.clawhunt.awwo.local.

• Updates and uninstalls keep it. The uninstaller removes it only if you ask it to, and asks again.
• Your canvases, runs and files are stored there.
• What a node sends to a model or media service goes to the provider whose key you added, or, through an AwwO 云端 connection, to the online AwwO.

数据保存在 %LOCALAPPDATA%\store.clawhunt.awwo.local。

• 升级和卸载都不会删除；卸载时可以选择删除，会再确认一次。
• 画布、运行记录和文件保存在这里。
• 节点发给模型或媒体服务的内容会发到你填入 Key 的服务商；通过 AwwO 云端连接时，发到线上 AwwO。

Install it inside your own user folder or %LOCALAPPDATA% (the default is), in a path of at most 80 characters without % or !. A Chinese user name or folder is fine. The AwwO menu restarts it, opens it in your browser, opens the data folder and the logs, and resets the sign-in password.
安装位置要在自己的用户目录或 %LOCALAPPDATA% 之内（默认位置就是），路径不超过 80 个字符，不含 % 或 !；用户名或路径里有中文也可以。菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg (and Chrome or Edge) installed. / 渲染成视频和剪辑节点需要自行安装 ffmpeg（以及 Chrome 或 Edge）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。

For a Mac with Apple silicon, AwwO Local is a separate download: AwwO for Mac 0.10.0. It has this update's two new how-tos and the 编辑 fix, not the rest yet.
Mac（Apple 芯片）版是单独的下载：AwwO for Mac 0.10.0。它已有本次更新中的两个新教程和「编辑」的修复，其余暂未包含。

Notes / 说明

The installer is not signed yet, so Windows may warn about an unknown publisher. It carries:

• every third-party licence (licenses/, build/pgsql/LIBRARIES.md);
• the source of the LGPL libraries libintl and libiconv (build/pgsql/sources/);
• beside PostgreSQL, Microsoft's redistributable Visual C++ runtime (14.44), from Visual Studio's redistributable folder.

安装包未签名，Windows 可能提示未知发布者。安装包内附带：

• 全部第三方许可证（licenses/、build/pgsql/LIBRARIES.md）；
• LGPL 库 libintl 与 libiconv 的源码（build/pgsql/sources/）；
• PostgreSQL 旁的微软可再分发 Visual C++ 运行库（14.44），取自 Visual Studio 的可再分发文件夹。

SHA-256 of AwwO-Local_x64-setup.exe: f1cc38f346dd6a2c34a27b840876a42404c40aae5fd9d05424714d17ecf3c834 (also in SHA256SUMS.txt).
Built from AwwO commit 067a32227c925563747327c3fe93cb425f0e3009.

AwwO 本地版 0.1.2 — AwwO Local for Windows

AwwO 本地版 0.1.2 — AwwO Local for Windows. Run AwwO on your own Windows computer: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits — and now, if you like, with your online AwwO account's models.

在自己的 Windows 电脑上运行 AwwO：AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度；现在也可以接入你线上 AwwO 账号的模型。

New in 0.1.2 / 0.1.2 更新

• AwwO 云端 (AwwO Cloud): your online account's models in AwwO Local.

  • On awwo.clawhunt.store, open 设置 → 本机版连接 and make a connection key for one of your workspaces. In AwwO Local, paste it under 「我的引擎」 → AwwO 云端.
  • Calls made through it run on the online AwwO's models that the workspace may use (Amazon Bedrock).
  • They count as that workspace's calls, on its plan and credits or its free allowance.
  • The online AwwO records each call's usage (the model, token counts, timings, an estimated cost and any error code), not its prompts or answers. See clause 08 of the Privacy Policy.
  • A key lasts 90 days, and you can revoke it on the same page at any time.
  • Your own keys under 「我的引擎」 keep working as before.

  AwwO 云端：在本地版里使用你线上 AwwO 账号的模型。

  • 在 awwo.clawhunt.store 打开「设置 → 本机版连接」，为你的一个工作区生成连接密钥，再粘贴到本地版的「我的引擎 → AwwO 云端」。
  • 通过它的调用在线上 AwwO 运行，使用该工作区可用的模型（Amazon Bedrock）。
  • 这些调用计入该工作区，按它的套餐和额度或免费额度计算。
  • 线上 AwwO 只记录每次调用的用量（模型、token 数、耗时、预估费用和错误码），不保存提示词和回答。详见隐私政策第 08 条。
  • 密钥有效 90 天，可以随时在同一页面撤销。
  • 你在「我的引擎」里填写的自己的 Key 照常可用。

• AwwO Local no longer takes a folder that holds other files as its data folder when the folder has no local.env: it refuses it and leaves it as it is.
  数据目录没有 local.env 却存有其他文件时，本地版不再把它当作自己的数据目录，而是拒绝并保持原样。

• If you have 0.1.1, install 0.1.2 over it: your data is kept.
  装过 0.1.1 的话，直接安装 0.1.2 覆盖即可，数据会保留。

Install / 安装

1. Download AwwO-Local_x64-setup.exe and run it. Windows 10 and 11, x64. It installs for the current account only and needs no administrator rights. It takes about 500 MB on disk, plus your data.
   下载 AwwO-Local_x64-setup.exe 并安装。适用于 Windows 10 和 11（x64）；只装给当前账号，不需要管理员权限；约占 500 MB 磁盘（另加你的数据）。
2. Open AwwO Local. Its first start sets up the database on this computer and opens admin-credentials.txt, which holds the first account's email and password.
   打开 AwwO Local。第一次启动会在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。
3. Sign in. Then add your own model API keys under 「我的引擎」 (My engines), or an AwwO 云端 connection key.
   登录后在「我的引擎」填写自己的模型 API Key，或填入 AwwO 云端的连接密钥。

Your data is in %LOCALAPPDATA%\store.clawhunt.awwo.local.

• Updates and uninstalls keep it. The uninstaller removes it only if you ask it to, and asks again.
• Your canvases, runs and files are stored there.
• What a node sends to a model or media service goes to the provider whose key you added, or, through an AwwO 云端 connection, to the online AwwO.

数据保存在 %LOCALAPPDATA%\store.clawhunt.awwo.local。

• 升级和卸载都不会删除；卸载时可以选择删除，会再确认一次。
• 画布、运行记录和文件保存在这里。
• 节点发给模型或媒体服务的内容会发到你填入 Key 的服务商；通过 AwwO 云端连接时，发到线上 AwwO。

Install it inside your own user folder or %LOCALAPPDATA% (the default is), in a path of at most 80 characters without % or !. A Chinese user name or folder is fine. The AwwO menu restarts it, opens it in your browser, opens the data folder and the logs, and resets the sign-in password.
安装位置要在自己的用户目录或 %LOCALAPPDATA% 之内（默认位置就是），路径不超过 80 个字符，不含 % 或 !；用户名或路径里有中文也可以。菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg (and Chrome or Edge) installed. / 渲染成视频和剪辑节点需要自行安装 ffmpeg（以及 Chrome 或 Edge）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。
• A macOS version comes later. / macOS 版稍后推出。

Notes / 说明

The installer is not signed yet, so Windows may warn about an unknown publisher. It carries:

• every third-party licence (licenses/, build/pgsql/LIBRARIES.md);
• the source of the LGPL libraries libintl and libiconv (build/pgsql/sources/);
• beside PostgreSQL, Microsoft's redistributable Visual C++ runtime (14.44), from Visual Studio's redistributable folder.

安装包未签名，Windows 可能提示未知发布者。安装包内附带：

• 全部第三方许可证（licenses/、build/pgsql/LIBRARIES.md）；
• LGPL 库 libintl 与 libiconv 的源码（build/pgsql/sources/）；
• PostgreSQL 旁的微软可再分发 Visual C++ 运行库（14.44），取自 Visual Studio 的可再分发文件夹。

SHA-256 of AwwO-Local_x64-setup.exe: 9dff963a9c1b24ba800cb74088655e212f2772c57d080d067061d0ada8a59d5a (also in SHA256SUMS.txt).
Built from AwwO commit cbcae908550c1e00f4ee62e8c365ac710d2ea12a.

AwwO 本地版 0.1.1 — AwwO Local for Windows

AwwO 本地版 0.1.1 — AwwO Local for Windows. Run AwwO on your own Windows computer: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits.

在自己的 Windows 电脑上运行 AwwO：AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度。

New in 0.1.1 / 0.1.1 更新

• Image and video generations run in your own workspace. 0.1.0 refused every one with model_not_allowed, though it offered the models. If you installed 0.1.0, install 0.1.1 over it: your data is kept.
  图片和视频生成可以在你自己的工作区运行了。0.1.0 虽然列出了这些模型，但每次生成都会被拒绝（model_not_allowed）。装过 0.1.0 的话，直接安装 0.1.1 覆盖即可，数据会保留。
• From this release on, the installer is always named AwwO-Local_x64-setup.exe, so the newest one is always at the same address.
  从这个版本起，安装包固定命名为 AwwO-Local_x64-setup.exe，最新版本的下载地址不变。

Install / 安装

1. Download AwwO-Local_x64-setup.exe and run it. Windows 10 and 11, x64. It installs for the current account only and needs no administrator rights; about 500 MB on disk, and your data.
   下载 AwwO-Local_x64-setup.exe 并安装。适用于 Windows 10 和 11（x64）；只装给当前账号，不需要管理员权限；约占 500 MB 磁盘（另加你的数据）。
2. Open AwwO Local. Its first start sets up the database on this computer and opens admin-credentials.txt, which holds the first account's email and password.
   打开 AwwO Local。第一次启动会在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。
3. Sign in and add your own model API keys under 「我的引擎」 (My engines).
   登录后在「我的引擎」填写自己的模型 API Key。

Your data is in %LOCALAPPDATA%\store.clawhunt.awwo.local. Updates and uninstalls keep it; the uninstaller removes it only if you ask it to, and asks again. Your canvases, runs and files are stored there; what a node sends to a model or media service goes to the provider whose key you added.
数据保存在 %LOCALAPPDATA%\store.clawhunt.awwo.local，升级和卸载都不会删除（卸载时可以选择删除，会再确认一次）。画布、运行记录和文件保存在这里；节点发给模型或媒体服务的内容会发到你填入 Key 的服务商。

Install it inside your own user folder or %LOCALAPPDATA% (the default is), in a path of at most 80 characters without % or !; a Chinese user name or folder is fine. The AwwO menu restarts it, opens it in your browser, opens the data folder and the logs, and resets the sign-in password.
安装位置要在自己的用户目录或 %LOCALAPPDATA% 之内（默认位置就是），路径不超过 80 个字符，不含 % 或 !；用户名或路径里有中文也可以。菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg (and Chrome or Edge) installed. / 渲染成视频和剪辑节点需要自行安装 ffmpeg（以及 Chrome 或 Edge）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。
• A macOS version comes later. / macOS 版稍后推出。

Notes / 说明

The installer is not signed yet, so Windows may warn about an unknown publisher. It carries every third-party licence (licenses/, build/pgsql/LIBRARIES.md) and the source of the LGPL libraries libintl and libiconv (build/pgsql/sources/). The Visual C++ runtime beside PostgreSQL is Microsoft's redistributable (14.44), from Visual Studio's redistributable folder.
安装包未签名，Windows 可能提示未知发布者。安装包内附带全部第三方许可证（licenses/、build/pgsql/LIBRARIES.md），以及 LGPL 库 libintl 与 libiconv 的源码（build/pgsql/sources/）。PostgreSQL 旁的 Visual C++ 运行库是微软的可再分发组件（14.44），取自 Visual Studio 的可再分发文件夹。

SHA-256 of AwwO-Local_x64-setup.exe: 7b411da1f26af6af88c81d0af824685bd2298ff2344385603203edd3b2bd95e2 (also in SHA256SUMS.txt).
Built from AwwO commit 720cf79dcc21bb7caa8dd12363a9fc3dade1dad9.

AwwO 本地版 0.1.0 — AwwO Local for Windows (superseded by 0.1.1)

Superseded by 0.1.1. 0.1.0 refuses every image and video generation in your own workspace (model_not_allowed). Install 0.1.1 over it: your data is kept.
已由 0.1.1 取代。 0.1.0 在你自己的工作区里会拒绝所有图片和视频生成（model_not_allowed）。直接安装 0.1.1 覆盖即可，数据会保留。

---

AwwO 本地版 0.1.0 — AwwO Local for Windows. Run AwwO on your own Windows computer: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits.

在自己的 Windows 电脑上运行 AwwO：AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度。

Install / 安装

1. Download AwwO-Local_0.1.0_x64-setup.exe and run it. Windows 10 and 11, x64. It installs for the current account only and needs no administrator rights; about 500 MB on disk, and your data.
   下载 AwwO-Local_0.1.0_x64-setup.exe 并安装。适用于 Windows 10 和 11（x64）；只装给当前账号，不需要管理员权限；约占 500 MB 磁盘（另加你的数据）。
2. Open AwwO Local. Its first start sets up the database on this computer and opens admin-credentials.txt, which holds the first account's email and password.
   打开 AwwO Local。第一次启动会在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。
3. Sign in and add your own model API keys under 「我的引擎」 (My engines).
   登录后在「我的引擎」填写自己的模型 API Key。

Your data is in %LOCALAPPDATA%\store.clawhunt.awwo.local. Updates and uninstalls keep it; the uninstaller removes it only if you ask it to, and asks again. Your canvases, runs and files are stored there; what a node sends to a model or media service goes to the provider whose key you added.
数据保存在 %LOCALAPPDATA%\store.clawhunt.awwo.local，升级和卸载都不会删除（卸载时可以选择删除，会再确认一次）。画布、运行记录和文件保存在这里；节点发给模型或媒体服务的内容会发到你填入 Key 的服务商。

Install it inside your own user folder or %LOCALAPPDATA% (the default is), in a path of at most 80 characters without % or !; a Chinese user name or folder is fine. The AwwO menu restarts it, opens it in your browser, opens the data folder and the logs, and resets the sign-in password.
安装位置要在自己的用户目录或 %LOCALAPPDATA% 之内（默认位置就是），路径不超过 80 个字符，不含 % 或 !；用户名或路径里有中文也可以。菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg (and Chrome or Edge) installed. / 渲染成视频和剪辑节点需要自行安装 ffmpeg（以及 Chrome 或 Edge）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。
• A macOS version comes later. / macOS 版稍后推出。

Notes / 说明

The installer is not signed yet, so Windows may warn about an unknown publisher. It carries every third-party licence (licenses/, build/pgsql/LIBRARIES.md) and the source of the LGPL libraries libintl and libiconv (build/pgsql/sources/). The Visual C++ runtime beside PostgreSQL is Microsoft's redistributable (14.44), from Visual Studio's redistributable folder.
安装包未签名，Windows 可能提示未知发布者。安装包内附带全部第三方许可证（licenses/、build/pgsql/LIBRARIES.md），以及 LGPL 库 libintl 与 libiconv 的源码（build/pgsql/sources/）。PostgreSQL 旁的 Visual C++ 运行库是微软的可再分发组件（14.44），取自 Visual Studio 的可再分发文件夹。

SHA-256 of AwwO-Local_0.1.0_x64-setup.exe: 000693fe648a22e50a41f4bb6a1f3fd5a970dca5f41cdfc552dbe1e95ff5b714 (also in SHA256SUMS.txt).
Built from AwwO commit 44ca8c77a75aeffb4616ed87d19c5bfc46f328cc.

AwwO for Mac 0.10.1 (build 12) — AwwO 本地版 for Mac

AwwO for Mac 0.10.1 (build 12), with AwwO 本地版 — AwwO Local for the Mac. One app for both of your AwwO workspaces: your online AwwO account, or AwwO Local on this Mac: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits — and, if you like, with your online AwwO account's models through AwwO 云端.

一个应用，两个工作区：线上工作区使用你的线上 AwwO 账号；本机工作区（AwwO 本地版）在这台 Mac 上运行 AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度；也可以通过 AwwO 云端接入你线上 AwwO 账号的模型。

New in 0.10.1 / 0.10.1 更新

In the local workspace (AwwO 本地版) / 本机工作区（AwwO 本地版）：

• The Academy's 从这里开始 (Start here) how-tos are numbered the same way on their title cards, 1 / 6 to 6 / 6.
  学院「从这里开始」的 6 个使用教程，片头统一编号为 1 / 6 到 6 / 6。
• Running the whole canvas asks first when nodes already have their result: run only the nodes without one (and what they feed), run every node again, or cancel.
  运行整张画布时，如果有节点已经有结果，会先询问：只运行还没有结果的节点（以及它们供给的节点）、全部重新运行，或者取消。
• A page that keeps data in localStorage, sessionStorage or cookies now works in the preview, which gives it an in-memory stand-in, instead of failing there and being repaired again and again.
  用 localStorage、sessionStorage 或 cookie 保存数据的页面现在可以在预览里正常运行（预览为它们提供内存中的替代），不再在预览中报错并被反复自动修复。
• Ask the canvas assistant to make one node depend on another and it wires them, instead of asking which field to use.
  让画布助手把一个节点设为依赖另一个节点时，它会直接连好，不再反问用哪个字段。
• When a model provider refuses a request, it is called too long for the model only when the provider's own error says so; a 422 answer counts as an ordinary model error.
  模型服务商拒绝请求时，只有服务商自己的错误信息表明超长，才判定为超出模型上下文；422 回复按普通模型错误处理。
• Seven built-in design skills, adapted from open-design and refero_skill, for nodes that deliver a page or a design: design craft, stripping the AI look, visual directions, UI states and the accessibility floor, motion discipline, a five-lens design jury, and procedural three.js scenes. Page nodes may load three.js r186 as ES modules.
  七个内置设计技能（改编自 open-design 与 refero_skill），供交付网页或设计的节点使用：设计工艺、去掉 AI 味、视觉方向库、界面状态与无障碍底线、动效纪律、设计评审团、程序化三维场景。页面节点可以用 ES 模块加载 three.js r186。
• A node that names the design-craft or anti-ai-slop skill, as the pages the online AwwO's canvas assistant plans now do, runs here too. Where those two would not fit the model's context, the run goes without them instead of being refused.
  带有「设计工艺」或「去掉 AI 味」技能的节点（线上 AwwO 的画布助手规划的页面现在会带上这两个技能）在本机工作区也能运行。这两个技能放不进模型上下文时，运行会去掉它们，而不是被拒绝。
• Those two skills accept the note they ask for in the hand-over as an HTML comment, so a page delivered as a single HTML file can carry it.
  这两个技能要求在交付说明里写的那句话，现在也可以写成 HTML 注释，单个 HTML 文件交付的页面也能带上。
• An edit's sound can be made by a page (its window.renderAudio), on the same frame clock as a film made as code.
  剪辑里的声音可以由页面生成（页面的 window.renderAudio），与用代码制作的影片使用同一套帧时钟。
• If you have 0.10.0, replace the app with 0.10.1: your local workspace's data is kept.
  装过 0.10.0 的话，直接用 0.10.1 替换应用即可，本机工作区的数据会保留。

What it is / 简介

• The first time it opens, it asks which workspace to open, and it remembers your choice. The 工作区 (Workspace) menu switches between them.

  • 线上工作区 (online workspace): your account, canvases and history at awwo.clawhunt.store.
  • 本机工作区 (local workspace, AwwO 本地版): everything stays on this Mac. It runs the same AwwO Local as 0.1.4 for Windows, built from the same commit.

• AwwO 云端 (AwwO Cloud) in 「我的引擎」 (My engines): on awwo.clawhunt.store, open 设置 → 本机版连接 and make a connection key for one of your workspaces, then paste it here. Calls made through it run on the online AwwO's models that the workspace may use (Amazon Bedrock) and count as that workspace's calls, on its plan and credits or its free allowance. The online AwwO records each call's usage (the model, token counts, timings, an estimated cost and any error code), not its prompts or answers; see clause 08 of the Privacy Policy. A key lasts 90 days and can be revoked on the same page. Your own keys keep working as before.

• 第一次打开时选择工作区，之后会记住；可以从菜单「工作区」切换。

  • 线上工作区：你在 awwo.clawhunt.store 的账号、画布和历史记录。
  • 本机工作区（AwwO 本地版）：一切都只在这台 Mac 上。它运行的本地版与 Windows 版 0.1.4 相同，由同一个提交构建。

• AwwO 云端（在「我的引擎」中）：在 awwo.clawhunt.store 打开「设置 → 本机版连接」，为你的一个工作区生成连接密钥，粘贴到这里。通过它的调用在线上 AwwO 运行，使用该工作区可用的模型（Amazon Bedrock），计入该工作区，按它的套餐和额度或免费额度计算。线上 AwwO 只记录每次调用的用量（模型、token 数、耗时、预估费用和错误码），不保存提示词和回答，详见隐私政策第 08 条。密钥有效 90 天，可以随时在同一页面撤销。你在「我的引擎」里填写的自己的 Key 照常可用。

Install / 安装

1. Download AwwO-macOS-0.10.1-build12-arm64.dmg, open it and drag AwwO Local into Applications. Apple silicon, macOS 14 or later. About 235 MB to download and 580 MB installed, plus your data.
   下载 AwwO-macOS-0.10.1-build12-arm64.dmg，打开后把 AwwO Local 拖进「应用程序」。适用于 Apple 芯片、macOS 14 或更新版本；下载约 235 MB，安装后约占 580 MB（另加你的数据）。
2. The app is signed ad hoc, not notarized, so the first time macOS says 「未打开“AwwO Local.app”」 (“AwwO Local.app” Not Opened). Choose 完成 (Done), then in System Settings → Privacy & Security choose 仍要打开 (Open Anyway) and confirm with Touch ID or your password; open the app again and choose 仍要打开.
   应用是 ad-hoc 签名，未经公证：第一次打开时 macOS 会提示「未打开“AwwO Local.app”」。点「完成」，再到「系统设置 → 隐私与安全性」点「仍要打开」，用触控 ID 或密码确认；再打开一次应用并选择「仍要打开」。
3. Choose 本机工作区 (AwwO 本地版). Its first start sets up the database on this Mac in a few seconds and opens admin-credentials.txt, which holds the first account's email and password. Sign in, then add your own model API keys under 「我的引擎」, or an AwwO 云端 connection key.
   选择「本机工作区（AwwO 本地版）」。第一次启动会在几秒内在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。登录后在「我的引擎」填写自己的模型 API Key，或填入 AwwO 云端的连接密钥。

Your local workspace's data is in ~/Library/Application Support/store.clawhunt.awwo.local.

• Replacing the app to update it keeps it, and so does deleting the app.
• Your canvases, runs and files are stored there.
• What a node sends to a model or media service goes to the provider whose key you added, or, through an AwwO 云端 connection, to the online AwwO.
• An earlier AwwO for Mac (0.9.x) had a local mode of its own: its folder ~/Library/Application Support/AwwO Local/ is never read or written. Quit the earlier app before you replace it.

本机工作区的数据保存在 ~/Library/Application Support/store.clawhunt.awwo.local。

• 替换应用升级、删除应用都不会删除它。
• 画布、运行记录和文件保存在这里。
• 节点发给模型或媒体服务的内容会发到你填入 Key 的服务商；通过 AwwO 云端连接时，发到线上 AwwO。
• 早期的 AwwO for Mac（0.9.x）有自己的本机模式，它的目录 ~/Library/Application Support/AwwO Local/ 从不被读写。替换前先退出旧应用。

The AwwO menu restarts AwwO Local, opens it in your browser, opens the data folder and the logs, and resets the sign-in password. Quitting the app stops AwwO Local in order.
菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码；退出应用时会有序停止本地版。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg and Google Chrome installed (for example brew install ffmpeg). / 渲染成视频和剪辑节点需要自行安装 ffmpeg 和 Google Chrome（例如 brew install ffmpeg）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。

Notes / 说明

The app is signed ad hoc: not with an Apple Developer ID, and not notarized. It carries:

• every third-party licence (Contents/Resources/payload/licenses/, build/pgsql/LIBRARIES.md);
• the GNU source of the LGPL library libiconv (build/pgsql/sources/).

应用为 ad-hoc 签名，没有 Apple Developer ID 签名，也未经公证。应用内附带：

• 全部第三方许可证（Contents/Resources/payload/licenses/、build/pgsql/LIBRARIES.md）；
• LGPL 库 libiconv 的 GNU 源码（build/pgsql/sources/）。

AwwO Local for Windows is AwwO 本地版 0.1.4. / Windows 版为 AwwO 本地版 0.1.4。

SHA-256 of AwwO-macOS-0.10.1-build12-arm64.dmg: e70efcc1a4e2ac3557ef81deaf35e0b2c0046c96c689712df728d02eeab8fdca (also in SHA256SUMS.txt).
Built from AwwO commit b60b7dec3638daddf66b7074676e888e4299cae3.

AwwO for Mac 0.10.0 (build 11) — AwwO 本地版 for Mac

AwwO for Mac 0.10.0 (build 11), with AwwO 本地版 — AwwO Local for the Mac. One app for both of your AwwO workspaces: your online AwwO account, or AwwO Local on this Mac: AwwO's own database, API, web app and agents, for you alone, with your own model keys, with no plan and no credits — and, if you like, with your online AwwO account's models through AwwO 云端.

一个应用，两个工作区：线上工作区使用你的线上 AwwO 账号；本机工作区（AwwO 本地版）在这台 Mac 上运行 AwwO 自己的数据库、API、网页和执行服务，只供你一人使用，用你自己的模型 Key，不需要套餐和额度；也可以通过 AwwO 云端接入你线上 AwwO 账号的模型。

What it is / 简介

• The first time it opens, it asks which workspace to open, and it remembers your choice. The 工作区 (Workspace) menu switches between them.

  • 线上工作区 (online workspace): your account, canvases and history at awwo.clawhunt.store.
  • 本机工作区 (local workspace, AwwO 本地版): everything stays on this Mac. It runs the same AwwO Local as 0.1.2 for Windows, plus the six later commits of AwwO's main branch; one of them stops 编辑 on a template guide from leaving Backspace deleting the node.

• AwwO 云端 (AwwO Cloud) in 「我的引擎」 (My engines): on awwo.clawhunt.store, open 设置 → 本机版连接 and make a connection key for one of your workspaces, then paste it here. Calls made through it run on the online AwwO's models that the workspace may use (Amazon Bedrock) and count as that workspace's calls, on its plan and credits or its free allowance. The online AwwO records each call's usage (the model, token counts, timings, an estimated cost and any error code), not its prompts or answers; see clause 08 of the Privacy Policy. A key lasts 90 days and can be revoked on the same page. Your own keys keep working as before.

• 第一次打开时选择工作区，之后会记住；可以从菜单「工作区」切换。

  • 线上工作区：你在 awwo.clawhunt.store 的账号、画布和历史记录。
  • 本机工作区（AwwO 本地版）：一切都只在这台 Mac 上。它运行的本地版与 Windows 版 0.1.2 相同，另加 AwwO 主分支之后的六个提交；其中一个修复了：在模板引导里按「编辑」后，Backspace 不再删除节点。

• AwwO 云端（在「我的引擎」中）：在 awwo.clawhunt.store 打开「设置 → 本机版连接」，为你的一个工作区生成连接密钥，粘贴到这里。通过它的调用在线上 AwwO 运行，使用该工作区可用的模型（Amazon Bedrock），计入该工作区，按它的套餐和额度或免费额度计算。线上 AwwO 只记录每次调用的用量（模型、token 数、耗时、预估费用和错误码），不保存提示词和回答，详见隐私政策第 08 条。密钥有效 90 天，可以随时在同一页面撤销。你在「我的引擎」里填写的自己的 Key 照常可用。

Install / 安装

1. Download AwwO-macOS-0.10.0-build11-arm64.dmg, open it and drag AwwO Local into Applications. Apple silicon, macOS 14 or later. About 235 MB to download and 580 MB installed, plus your data.
   下载 AwwO-macOS-0.10.0-build11-arm64.dmg，打开后把 AwwO Local 拖进「应用程序」。适用于 Apple 芯片、macOS 14 或更新版本；下载约 235 MB，安装后约占 580 MB（另加你的数据）。
2. The app is signed ad hoc, not notarized, so the first time macOS says 「未打开“AwwO Local.app”」 (“AwwO Local.app” Not Opened). Choose 完成 (Done), then in System Settings → Privacy & Security choose 仍要打开 (Open Anyway) and confirm with Touch ID or your password; open the app again and choose 仍要打开.
   应用是 ad-hoc 签名，未经公证：第一次打开时 macOS 会提示「未打开“AwwO Local.app”」。点「完成」，再到「系统设置 → 隐私与安全性」点「仍要打开」，用触控 ID 或密码确认；再打开一次应用并选择「仍要打开」。
3. Choose 本机工作区 (AwwO 本地版). Its first start sets up the database on this Mac in a few seconds and opens admin-credentials.txt, which holds the first account's email and password. Sign in, then add your own model API keys under 「我的引擎」, or an AwwO 云端 connection key.
   选择「本机工作区（AwwO 本地版）」。第一次启动会在几秒内在本机建立数据库，并打开 admin-credentials.txt：里面是初始账号和密码。登录后在「我的引擎」填写自己的模型 API Key，或填入 AwwO 云端的连接密钥。

Your local workspace's data is in ~/Library/Application Support/store.clawhunt.awwo.local.

• Replacing the app to update it keeps it, and so does deleting the app.
• Your canvases, runs and files are stored there.
• What a node sends to a model or media service goes to the provider whose key you added, or, through an AwwO 云端 connection, to the online AwwO.
• An earlier AwwO for Mac (0.9.x) had a local mode of its own: its folder ~/Library/Application Support/AwwO Local/ is never read or written. Quit the earlier app before you replace it.

本机工作区的数据保存在 ~/Library/Application Support/store.clawhunt.awwo.local。

• 替换应用升级、删除应用都不会删除它。
• 画布、运行记录和文件保存在这里。
• 节点发给模型或媒体服务的内容会发到你填入 Key 的服务商；通过 AwwO 云端连接时，发到线上 AwwO。
• 早期的 AwwO for Mac（0.9.x）有自己的本机模式，它的目录 ~/Library/Application Support/AwwO Local/ 从不被读写。替换前先退出旧应用。

The AwwO menu restarts AwwO Local, opens it in your browser, opens the data folder and the logs, and resets the sign-in password. Quitting the app stops AwwO Local in order.
菜单「AwwO」可以重新启动、在浏览器中打开、打开数据目录与日志、重置登录密码；退出应用时会有序停止本地版。

Not included yet / 暂不包含

• Direct execution: its workspace image can only be built from the code repository for now. / 直接执行：它的工作区镜像目前只能在代码库中构建。
• Sandbox computers: they take only platform models, not your own keys. / 沙箱电脑：它只接平台模型，不接个人 Key。
• The Python OpenAI Agents worker: AwwO Local uses the JavaScript one. / Python 版 OpenAI Agents 执行器（使用 JavaScript 版）。
• Rendering to video and the clip nodes need ffmpeg and Google Chrome installed (for example brew install ffmpeg). / 渲染成视频和剪辑节点需要自行安装 ffmpeg 和 Google Chrome（例如 brew install ffmpeg）。
• The Academy's videos play from the hosted site. / 学院视频从线上站点播放。

Notes / 说明

The app is signed ad hoc: not with an Apple Developer ID, and not notarized. It carries:

• every third-party licence (Contents/Resources/payload/licenses/, build/pgsql/LIBRARIES.md);
• the GNU source of the LGPL library libiconv (build/pgsql/sources/).

应用为 ad-hoc 签名，没有 Apple Developer ID 签名，也未经公证。应用内附带：

• 全部第三方许可证（Contents/Resources/payload/licenses/、build/pgsql/LIBRARIES.md）；
• LGPL 库 libiconv 的 GNU 源码（build/pgsql/sources/）。

AwwO Local for Windows is now AwwO 本地版 0.1.4. / Windows 版现为 AwwO 本地版 0.1.4。

SHA-256 of AwwO-macOS-0.10.0-build11-arm64.dmg: 5ead34849a3a4dd3b5ac717e59cf80353dd0aae20abf1b9d7bc7e47306c4b72a (also in SHA256SUMS.txt).
Built from AwwO commit 68e7923462bb6788f24c68786866e29c640013e3.