Introduction

The DSH community plugin ecosystem emphasizes “everything is a plugin.” The community directory is an independent site with no official affiliation to DeepSeek / Huanfang. @dely0/dsh-personal-workbench is a DSH Web plugin maintained by Dely0 that turns DSH Web into a personal workbench: a calendar, a tree-structured task list, and an AI assistant revolving around tasks.

Below is an introduction to its capabilities, installation methods, typical usage, and usage notes.

What is this

One-sentence positioning: @dely0/dsh-personal-workbench is a DSH personal workbench plugin that provides a calendar + task list + AI assistant workbench.

It is suitable for local single-user scenarios:

  • Task data is stored locally in ~/.dsh/workbench.
  • The documentation mentions an SQLite database file ~/.dsh/workbench/workbench.db and mentions a daily JSON backup plan.
  • The workbench API is mounted on /api/workbench/* and only allows loopback access.
  • AI capabilities depend on the models and credentials you have configured in DSH; operations like Execute / Consult will actually consume tokens.
  • The license is MIT.

Core Features

Calendar and Task List

  • Calendar supports week / month switching.
  • Task list is tree-structured.
  • Supports Markdown task descriptions, review records, and change history.

AI Task Workflow

  • Quick entry via natural language, automatically generating tasks after AI clarification.
  • Each task can be associated with multiple AI sessions: Clarify / Consult / Break down / Execute / Review.
  • Task execution uses a closed loop of “AI requests completion → User acceptance.”
  • Completed tasks can trigger AI review.

Reminders, Archiving, and Recurring Tasks

  • Expiration reminders: in-page banners.
  • System-level desktop notifications.
  • Archive area, task recovery.
  • Recurring tasks.

Sorting, Reports, Knowledge Base, and Ideas

  • AI smart sorting, supports arbitrary dates.
  • AI daily report / weekly report.
  • Personal knowledge base / Error collection.
  • Knowledge base enhancement: AI summarizes local documents + file links.
  • Ideas / Idea King.

Installation and Enablement

Install from npm

Common installation commands:

dsh plugin --profile web add @dely0/dsh-personal-workbench

You can also install to a project using npm:

npm install @dely0/dsh-personal-workbench

Install from GitHub

dsh plugin --profile web add git+https://github.com/Dely0/dsh-personal-workbench.git

Mount from source code

If you need local modifications, you can clone the repository, install dependencies, check, test, build, and then mount:

git clone https://github.com/Dely0/dsh-personal-workbench.git
cd dsh-personal-workbench
pnpm install
pnpm check
pnpm test
pnpm build
dsh plugin --profile web add link:/path/to/dsh-personal-workbench

Restart and Refresh

After installation, restart dsh web, then perform a hard refresh in the browser:

Ctrl+Shift+R

Typical Usage

The following steps are used to understand the plugin’s main workflow.

  1. Install the plugin and restart dsh web.
  2. Hard refresh the page in the browser.
  3. Enter tasks using natural language and wait for the AI to clarify and generate tasks.
  4. Initiate AI consultation, breakdown, execution, or review for the task.
  5. For execution-type tasks, the AI submits a completion request, and the task is only considered complete after user acceptance.
  6. Use AI smart sorting to view execution order proposals for arbitrary dates.
  7. Use AI daily report / weekly report to generate periodic reports.
  8. Deposit experience, notes, or decisions into the personal knowledge base / error collection.
  9. Use knowledge base enhancement for local documents to let AI summarize the content and retain file links.
  10. Use Ideas / Idea King for inspiration, which can be converted to tasks later.

If inconsistent completion statuses of parent tasks appear in historical data, you can run:

pnpm repair

Or call:

POST /api/workbench/maintenance/repair-parents

Used to idempotently fill in historical parent task completion statuses.

Applicable Scenarios and Notes

Who is it for

  • Local single users who want to put their calendar, task list, and AI sessions in the same workbench.
  • People who have already configured models and credentials in DSH.
  • People who want data to stay locally in ~/.dsh/workbench and not rely on cloud sync.

Limitations

  • Only supports local single-user usage.
  • No cloud sync, no multi-user permission system.
  • The current version is developed and tested for DSH 0.1.0-rc.6 Web version.
  • When DSH upgrades to a new major version, selectors must be re-verified, and adaptation is necessary if needed.
  • When coexisting with dsh-web-ui (task-board / ssh), use its data-dsh-* mutual exclusion protocol; it automatically fails when not installed and does not depend on dsh-web-ui.

Security and License

  • Data is completely stored locally in ~/.dsh/workbench and not uploaded to any server.
  • The documentation mentions that it does not read or upload any data outside of DSH.
  • All workbench APIs are mounted on /api/workbench/* and only allow loopback access.
  • AI capabilities depend on the models and credentials you have configured in DSH; operations like Execute / Consult will actually consume tokens.
  • This plugin is a community project, unrelated to DeepSeek officially, and provides no guarantees.
  • It will run on the local machine with the current DSH user permissions.
  • Execution-type AI operations may modify workspace files and consume API quotas. Please read the code carefully before use.
  • The license is MIT.

Conclusion

The value of @dely0/dsh-personal-workbench lies in putting calendar, tasks, AI sessions, and local knowledge accumulation into the same DSH Web workbench, suitable for local single users to connect task flows with AI workflows.

GitHub Repository:

https://github.com/Dely0/dsh-personal-workbench

The directory page address is not listed in this article; it is recommended to check the README, code, and MIT license in the repository before installing.