Introduction

The core design philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” When developing or demonstrating agent applications, a standardized data display or business simulation environment is often required. The guannan1031/dsh-commerce-cockpit plugin is designed for this purpose. It provides a runnable e-commerce operations cockpit and supports two modes: Demo data and CSV data import, addressing the need to build an e-commerce data dashboard from scratch.

Plugin Overview

This is a DeepSeek Harness client plugin maintained by guannan1031. It is mainly used to display business dashboards, perform data integrity checks, and provide rule-based data querying capabilities. The plugin is currently published on GitHub and uses the MIT license.

Core Features

The plugin includes the following verified features:

  1. Business Overview: displays sales revenue, ad spend, order count, conversion rate, average order value, and trends.
  2. Mode support:
    - Demo mode: uses a built-in seed snapshot and supports estimated operating contribution profit, simulated inventory, and a demonstration action list.
    - Imported mode: reads local CSV files and displays the metrics actually provided by the CSV.
  3. Data integrity checks: identifies missing fields and non-computable items.
  4. Markdown export: generates a one-page business brief.
  5. Rule-based data querying: supports trend analysis, channel comparison, and ad spend output ratio calculation through the cockpit_ask interface.

Installation and Enablement

Install it using the official DeepSeek Harness plugin mechanism:

dsh plugin --profile web add @guannan1031/dsh-commerce-cockpit@0.1.2
dsh plugin --profile desktop add @guannan1031/dsh-commerce-cockpit@0.1.2

After installation, restart the corresponding Harness profile and locate the “Cockpit” entry at the top of the conversation to use it.

Typical Usage

The plugin provides two running modes. Choose one based on your data source.

Demo Mode

This is the default mode and requires no additional configuration. The system uses a fixed seed snapshot, making it suitable for quickly reviewing dashboard behavior and testing cockpit_ask rules.

Imported Mode

Place CSV files in the $DSH_HOME/imports/commerce-cockpit/ directory (or ~/.dsh if DSH_HOME is not set). The plugin only reads files from that directory and does not fall back to Demo data.

CSV Field Requirements

Files must be UTF-8 encoded, with a maximum size of 5 MB and a maximum of 50000 rows. The minimum field set is as follows:

business_date,platform,store_id,channel,gmv,orders,visitors,ad_spend

The unique key is business_date + platform + store_id + channel. Note: null values and 0 are treated differently. When orders are 0, average order value cannot be calculated; when visitors are 0, conversion rate cannot be calculated; and when promotion spend is 0, the overall ad spend output ratio cannot be calculated.

Use Cases and Considerations

When using this plugin, pay attention to the following facts and limitations:

  1. Metric definitions:
    - The “estimated operating contribution profit” in Demo mode is a demonstration formula and does not include real refund, storage, or logistics metrics.
    - The “overall ad spend output ratio” only represents GMV / promotion spend and does not represent ad-attribution ROI.
    - Imported mode only answers questions supported by source data; if required fields are missing, it explicitly declines to answer.
  2. Feature boundaries:
    - It does not connect to real e-commerce platforms and has no API calls, automated task execution, real inventory, or competitor monitoring.
    - Inventory and competitor information in Demo mode are simulated only.
    - It does not process consumer personal information.

Summary

This plugin is a feature-complete demonstration layer and is suitable for developers building a baseline environment for e-commerce data analysis. For more details, refer to the project directory or the GitHub repository.