记事喵 lets users capture Chinese text or voice messages in WeChat and convert them into dated, grouped reminders, while keeping task data in Apple’s ecosystem and supporting local AI models.
A solo developer has built a Mac-first task capture tool called 记事喵, aimed at removing the friction between a passing thought and a properly scheduled reminder. The idea is simple: instead of sending notes to WeChat’s File Transfer Assistant and hoping they are copied into a to-do app later, the user can message the app directly in Chinese, by text or voice, and have the task parsed into a date, a group and a reminder automatically. The same item then appears in the Mac menu bar and in iPhone Reminders, with alerts firing at the appointed time.
The workflow is designed to suit both mobile and desk use. On the phone, the app connects to WeChat through Tencent’s official iLink protocol after the user taps “connect WeChat” and confirms by scanning a code. On the Mac, a global shortcut, Option-Command-T, opens a quick entry box from any app, so a task such as “明天下午三点前交周报” can be entered in seconds. The app also accepts follow-up commands in the same conversation, including queries about the week’s workload, moving a task to the following Monday, or marking an item complete.
The developer says the system stores no server-side data. Tasks live in Apple’s Reminders database, which syncs across devices through the user’s Apple account, so the Mac can remain switched off without affecting alerts. That local-first approach places the product in a growing group of lightweight productivity tools that avoid central account systems and keep user data on-device or in iCloud. Itsypad, for example, offers a similarly privacy-focused scratchpad with iCloud synchronisation, while Firetask and TaskPort take a more conventional cross-device task-manager route with iCloud-backed sync and native Apple-platform clients.
The app is also built around user-controlled AI rather than a fixed cloud dependency. According to the developer, it can work with an OpenAI-compatible endpoint or a local Ollama model, and only the task title, status and grouping are sent to the model, not the note field. If the model is uncertain, it asks for confirmation; the app then echoes back the understood deadline so the user can correct mistakes immediately. Deleted items can be restored, and the developer says the macOS build is native Swift, weighs just over 3 MB and is released under Apache 2.0. The current release requires macOS 14 or later on Apple Silicon, and the first launch must be allowed manually in macOS privacy and security settings because it has not been notarised.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
