Introduction

The philosophy of DSH is “everything is a plugin.” Work that originally required external tools can also be processed within the dsh web workflow. Interview practice is often scattered across question banks, chat logs, resume documents, and algorithm websites, making it difficult to quickly review, compare, and continue after answering.

dsh-interview is an AI interview practice assistant running in DeepSeek Harness Web, maintained by codingayice and licensed under MIT. It integrates questioning, answering, scoring and commenting, knowledge explanation, ready-to-speak answers, and the saving and viewing of practice records into a single flow.

What is this

dsh-interview can be understood as a local-first interview training workspace within DSH Web:

  • Supports five practice modes: memorizing common interview questions, mock interviews, resume-based prediction, scenario questions, and brushing LeetCode (LeetCode Hot 100).
  • Supports questioning, answering, scoring and commenting, knowledge explanation, and providing ready-to-speak answers.
  • Automatically saves questions, answers, scores, comments, explanations, and practice summaries.
  • Provides a visual practice workspace, supporting search, filter, switch, edit, delete, and export of practices.
  • Practice data is saved locally by default, requiring no extra database or service deployment.

Core Capabilities

Five Practice Modes

Here is an introduction to the scenarios suitable for each of the five modes.

Memorizing Common Questions

Suitable for focused review of a specific knowledge topic. You can fill in the topic to review, for example:

  • Java Concurrency
  • JVM
  • MySQL
  • Redis
  • Spring
  • Computer Networks

Mock Interview

Suitable for practicing around real interview scenarios. You can fill in your resume, target job, interviewer style, and interview difficulty, and choose whether to provide the JD and whether to code without an IDE.

Resume-based Prediction

Suitable for preparing in advance based on your resume and job requirements before the interview. You can fill in your resume, target job, whether to provide JD, the prediction scope, and interview difficulty.

Scenario Questions

Suitable for practicing open-ended questions such as system design, troubleshooting, performance issues, and engineering decisions. You can fill in the topic you want to practice, for example:

  • Online interface suddenly slows down
  • Design a seckill system
  • How to troubleshoot message backlog
  • How to handle database performance degradation

LeetCode Practice

Suitable for continuous algorithm practice. You can choose C++, Java, Python, C, or Go, draw questions from LeetCode Hot 100, and view algorithm explanations and complete code.

Practice Workflow

In supported coaching practices, a single question typically goes through the following process:

提问 → 作答 → 评分与点评 → 知识讲解 → 可直接表达的答案

After completing one question, you can continue to the next, or view the explanation and answer. Retaking the question will not overwrite previous records, making it easy to compare changes in multiple answers.

Practice Records

dsh-interview automatically saves:

  • Question
  • Answer
  • Score
  • Comment
  • Explanation
  • Practice Summary

Through the saving process above, you can continue unfinished practices, view history, compare multiple answers, and manage existing practices in the workspace.

Visual Practice Workspace

The workspace allows you to view:

  • In-progress practices
  • Finished practice archives
  • LeetCode Hot 100 questions and completion status

It also supports search, filter, switch, edit, delete, and export of practices.

Installation and Activation

First, confirm that your runtime environment meets the requirements:

  • DeepSeek Harness Web
  • Node.js 22.5 or higher

Install the plugin:

dsh plugin --profile web add dsh-interview

After installation, restart dsh web:

dsh web

If you need to update the plugin later, you can use:

dsh plugin --profile web update dsh-interview

Typical Usage

Create Practice

  1. Open the sidebar and click “Interview Training” above the settings.
  2. Click “New Practice”.
  3. Select a practice mode and fill in the corresponding content.
  4. Click “Start Practice”.
  5. Answer the question in the chat; supports a mode to view answers, or you can directly view the explanation and answer.

If you don’t want to manually fill in the configuration, you can also input in the chat:

新建练习

The AI will guide you through completing the practice configuration.

LeetCode Practice Example

Choose a language, for example:

  • C++
  • Java
  • Python
  • C
  • Go

The plugin will draw questions from LeetCode Hot 100 and display algorithm explanations and full code in the selected language.

Applicable Scenarios and Notes

This plugin is suitable for people preparing for technology development interviews, those wanting to review common interview questions, those wanting to predict questions based on their resume, those wanting to practice scenario questions, or those wanting to brush LeetCode Hot 100 and keep completion records.

Please pay attention to the following points before use:

  • Questioning, commenting, explaining, and normal practice summaries require AI.
  • Practice data is saved locally by default, requiring no extra database or service deployment.
  • Real resumes and answers are not automatically uploaded to other systems.
  • dsh-interview is an MIT open source project.
  • The plugin runs with the permissions of the current dsh process; check the source code and license before installing.
  • Requires Node.js 22.5 or higher.

Links

  • GitHub: https://github.com/codingayice/dsh-interview
  • Directory page (Plugin clue provided): https://www.skillhub.cn/plugins/codingayice/dsh-interview