kevin-qiao/dsh-todo: Add a sticky note to-do wall to the DSH Web interface.

Introduction

When developing DSH agent applications, tasks often scatter across conversations: requirements noted on the fly, clues to handle in the next round, points to review later. DSH’s philosophy is “everything is a plugin,” and some members of the community have also moved tasks directly into the Web GUI. Below introduces kevin-qiao/dsh-todo: a bilingual sticky note todo panel that appears in the left sidebar after installation, supporting both a sticker board and a minimalist list view, using AI to schedule, categorize, and provide suggestions one by one, with zero runtime dependencies.

Compared to listing tasks directly in the conversation, it deposits tasks into the interface and local files, preserved across sessions. Compared to external task management tools, it doesn’t introduce new services; ready to use after installation. It is suitable for individuals to maintain a lightweight todo list within DSH.

What is this

One-sentence positioning: A DSH Web Plugin that adds a sticky note todo wall to the left sidebar, bilingual, AI helps you schedule, categorize, discuss one by one, and remember habits.

Basic information:

  • Author / Maintainer: kevin-qiao
  • Repository: https://github.com/kevin-qiao/dsh-todo
  • Directory Page: https://www.skillhub.cn/plugins/kevin-qiao/dsh-todo
  • Category: Client-side
  • Version: 1.8.0 (README badge)
  • License: MIT
  • Runtime Dependencies: Zero

It solves three specific problems: tasks need long-term storage and reminders; a single task may need to be broken down into several steps; completion and rescheduling need historical records. The plugin puts these three things into the left sidebar and exposes tasks to the Agent via the sticker_todo tool.

Core Features

The following only lists capabilities explicitly stated in the README.

Two views:

  1. Sticker Board (Default)
  2. Minimalist List (shows only titles, click to expand details)

New Popup Fields:

  • Title + Details
  • Color
  • Priority (visual cue only, no forced sorting)
  • Quick Due Dates: Today / Tomorrow / The Day After Tomorrow / Next Week
  • Scheduled Reminders
  • Up to 3 tags

Organization:

  • Filter by Tag
  • Group by Tag
  • Completed Notes Area

Interface:

  • Font Presets (Chinese Handwriting / Xingkai / UI Font)
  • Language: Auto / Chinese / English
  • Smooth scroll animation, respecting system prefers-reduced-motion

AI Capabilities:

  1. Automatic Scheduling
  2. Categorization and Tagging – processes only notes without tags
  3. Single Note Coach: offers 2–4 selectable answers first, up to 3 rounds, suggestions are editable

Memory Capabilities:

  • History of rescheduling and completion
  • Editable preferences
  • Weekly Review
  • Coach conversation continues across sessions
  • Display “Rescheduled N times” if rescheduled more than twice

Hand over to DSH: The “Hand over to DSH” button on the note constructs a prompt with title, details, deadline, priority, tags, and id, puts it into the main dialogue draft, and does not auto-send.

Subcommands of the Agent tool sticker_todo:

list / add / update / complete / remove / clear_completed / plan / advice / tag / remember / forget / memories / review

Installation and Enablement

Installation command taken from README (PowerShell):

dsh plugin --profile web add dsh-sticker-todo

Restart DSH after installation. The entry point is “Sticker Todo” in the left sidebar; data is stored locally.

README also provides a way to install from source: clone the repository first, then use link to mount:

git clone https://github.com/kevin-qiao/dsh-todo.git
dsh plugin --profile web add link:D:\path\to\dsh-todo

Replace the path in the second command with the actual local location.

Typical Usage

Common header operations:

  • New Note
  • Font
  • Language (Auto / Chinese / English)
  • Sticker / Minimalist view toggle
  • Memory
  • Refresh

Note actions:

  • Coach Suggestion
  • Hand over to DSH
  • Remind in 1 hour
  • Edit, two-step delete

AI cards are used for automatic scheduling, categorization/tagging, and clearing schedules.

Memory Panel: Manage preferences, generate weekly review in “Memory”; view AI proposals (only saved to disk when clicked “Remember”); “Clear Memory” only clears history, leaving notes untouched.

Data and Fallback

Default data file path, overridable via dataFile config:

~/.dsh-sticker-todo/todos.json

Atomic write on every modification. Document v2 structure:

{ "version": 2, "todos": [], "events": [], "memory": { "facts": [], "threads": [] } }

Old files migrate automatically; old versions only read todos; downgrade won’t lose notes. When AI is unavailable (no model configured, timeout, non-compliant output), everything falls back to local rules and is marked source: "fallback", without errors. Deleting the data file will reset everything including notes; “Clear Memory” only clears events, preferences, and conversations.

Applicable Scenarios and Notes

Suitable for maintaining personal lightweight tasks in the DSH Web GUI, for those who need AI organization and review, and for developers wanting to expose a unified todo tool to the Agent. Zero runtime dependencies, Node ≥ 20, CI runs three sets of validation scripts on Node 20 / 22 / 24.

Notes:

  • The plugin runs with the current dsh process permissions; check source code and license before installation (this plugin is MIT).
  • There is only one local JSON for todos; deleting the file resets it, “Clear Memory” does not.
  • The directory page is an independent community site with no official affiliation to DeepSeek / Hypersphere, and is not treated as an official app store.
  • This plugin has 0 stars on the directory page, the ecosystem is relatively new, so evaluation is based on source code and documentation.

Summary

The value of dsh-todo lies in isolating tasks from the conversation: a sticky note wall in the left sidebar, plus AI scheduling, categorization, step-by-step coaching, and cross-session memory, combined with the sticker_todo tool to integrate with the Agent, with zero runtime dependencies. One command to install, data lands in local JSON.

Directory Page: https://www.skillhub.cn/plugins/kevin-qiao/dsh-todo

GitHub: https://github.com/kevin-qiao/dsh-todo

羽毛球分组比赛记分
小程序二维码

欢迎使用《羽毛球分组比赛记分》微信小程序

Xiaoye