cindy
Cindy is an open-source AI agent that runs locally on your machine, integrates multiple AI harnesses and models, and provides memory, skills, and automation to perform real work in your projects and apps.
它通过巧妙的变通方法(如使用不可见覆盖窗口和轮询)解决了此前无法完成的任务——提取并与小部件内容交互,从而使小部件专属数据可用于自动化。
Read and click Android home screen widgets from Tasker.
Lots of apps show information in a widget that they expose nowhere else — no API, no notification, no content provider. WidgetRelay hosts those widgets invisibly in the background, extracts their contents, and hands them to Tasker as plain text or JSON. It can also click elements inside them, so a widget button becomes something a task can trigger.
Typical uses:
WidgetRelay registers four Tasker plugin actions. All of them are configured the same way: pick an app and widget, choose a size, and — for all but the JSON action — tap the element you want in the extracted-data list.
| Action | Input | Output |
|---|---|---|
| Get Widget Data | element path | %widget_value — the element's text, content description or resource id |
| Get Widget JSON | — | %widget_json — the whole widget as a JSON tree |
| Click Widget Path | element path | — |
| Click Widget Text | element text | — |
Elements are addressed by their path in the widget's view tree: /root is
the widget itself, /root/0/2 is the third child of the first child, and so on.
The configuration screen shows every extracted element with its path, so you
normally never type one by hand — you tap the element and the path is filled in.
Paths are stable as long as the widget's layout doesn't change. If a widget
rearranges itself (many do, depending on state or size), prefer Click Widget
Text, which searches by the visible text or content description instead. Both
input fields accept Tasker variables, so %par1 or %text work too.
Get Widget JSON returns the widget as a nested tree. type is the view
class, children is only present when a node ha
Read and click Android home screen widgets from Tasker.
根据分类、Topic 和编程语言匹配的相似项目。
Cindy is an open-source AI agent that runs locally on your machine, integrates multiple AI harnesses and models, and provides memory, skills, and automation to perform real work in your projects and apps.
AetherST Tunnel is an Android VPN client designed for high-performance censorship circumvention, using advanced protocols like MASQUE, WireGuard, and nested WireGuard (Gool) to evade Deep Packet Inspection.
Nativ is a native macOS app that lets you run AI models locally on Apple silicon, offering chat, model management, performance analytics, and an OpenAI/Anthropic-compatible API server.