Introduction¶

In the plugin scenario of DeepSeek Harness (hereinafter referred to as DSH), a common type of requirement is not direct trading, but rule research centered around a single ticker: checking the current rule status, backtesting historical performance, generating interactive reports, and saving verified rules for reuse. If these capabilities are split into data fetching, Python backtesting, HTML charts, and strategy management, developers often need to maintain their own external toolchain.

dsh-quant-workspace offers another path: as a self-contained DSH quantitative research plugin, it features a built-in Python engine, supporting Yahoo Finance daily data, rule strategy backtesting, and interactive visualization reports. Below is an introduction to its positioning, core capabilities, installation method, and typical usage.

What is it¶

AllenCX/dsh-quant-workspace is a self-contained quantitative research workspace for DSH. It integrates data fetching, rule backtesting, and report generation into a single plugin, and provides the single_ticker tool for single-ticker research.

The project repository is AllenCX/dsh-quant-workspace and the license is MIT. It requires DSH’s web profile and pnpm; the built-in Python engine requires uv to run and syncs python/.venv on first use; fetching Yahoo Finance data requires internet access. The Node engine requirement is >=22.19.

Core Features¶

Data Acquisition¶

The plugin supports fetching Yahoo Finance daily data, defaulting to 2 years of data. It can export OHLCV data and indicator data.

Rule Backtesting¶

The backtesting executes against rule strategies, and the output content includes:

• Trade-by-trade table
• Total Profit
• Maximum Drawdown
• Win Rate
• Average Position Size
• Buy-and-Hold Baseline

Interactive Reports¶

The plugin can generate self-contained interactive HTML reports. The report content includes:

• Candlestick bands
• Volume
• %B
• Equity Curve

The report supports zooming, panning, crosshair, and toggling between red-up and green-down color switching. The HTML file is in a self-contained format and can be opened in a browser.

Strategy Registry¶

The plugin provides a strategy registry for saving researched strategies and reusing them by id. This way, rule parameters do not need to be re-described in subsequent sessions.

Read-only Boundary¶

The plugin is designed as read-only: it does not place orders, does not change positions, and does not require a market data key. It displays rule status and evidence, but the decision is the user’s responsibility and does not constitute investment advice.

Installation and Activation¶

Prerequisites¶

The following conditions must be met before use:

• DSH is installed and using the web profile
• pnpm is installed
• uv is installed to run the built-in Python engine
• Internet access is available to fetch Yahoo Finance data

Installation¶

The README indicates that this plugin has not yet been published to npm. Before publication, you can use the following GitHub installation command:

dsh plugin --profile web add github:AllenCX/dsh-quant-workspace

Another installation command also appears in the README:

dsh plugin --profile web add dsh-quant-workspace

These two commands appear simultaneously in the README; the officially recommended command needs to be confirmed based on the specific documentation and usage environment.

Configuration¶

All configuration items are optional. An example user patch path is as follows:

$DSH_HOME/profiles/web/cordis.patch.yml

Example content:

- id: quant-workspace
  config:
    ledgerPath: 'C:\path\to\trade_log.csv'   # optional: track your real positions
    reportsDir: 'C:\path\to\reports'          # optional: where visual reports go

Main configuration items are as follows:

Typical Usage¶

Usage in DSH Sessions¶

In a Harness session, you can trigger the single_ticker tool directly using natural language.

View today’s signal card for a ticker:

Give me today's signal card for TSLA

This corresponds to single_ticker, with mode set to daily.

Backtest a ticker and generate a chart:

Backtest META and generate a chart

This corresponds to single_ticker, with mode set to backtest, and chart: true enabled.

Perform a review health check:

Compare the review health check for TSLA

This corresponds to single_ticker, with mode set to review.

A single invocation runs only one rule. The rule can come from the example rule family or from a strategy id in the registry; if neither is provided, the defaultRule from the configuration is used. If none of these are provided, the workspace will report that no strategy is selected.

single_ticker Parameters¶

The main parameters for single_ticker are as follows:

• ticker: Required, ticker symbol, e.g., TSLA; automatically converted to uppercase, allows letters, numbers, dots, and hyphens only
• mode: Defaults to daily, optional values are daily, backtest, review
• rule: Example rule family, currently includes bollinger_mean_reversion
• strategy: Strategy id from the strategy registry
• chart: Defaults to false; when set to true, an interactive HTML report is generated simultaneously, a status CSV is written under reportsDir, and the output includes the path to the artifacts

CLI Usage¶

The plugin provides the dsh-quant CLI. The single-ticker command format is as follows:

dsh-quant single-ticker --ticker <T> --mode <daily|backtest|review> (--rule <family> | --strategy <id>) [--ledger <path>] [--chart <dir>] [--export-state <dir>] [--registry <path>] [--data-file <csv>]

Example commands for the strategy registry:

dsh-quant strategy register --id tsla_dip --family bollinger_mean_reversion --bollinger-window 30 --note 'dip strategy after Aug-2026 research'

View registered strategies:

dsh-quant strategy list

Remove a strategy:

dsh-quant strategy remove --id tsla_dip

CLI exit codes are as follows:

• 0: Success, outputs plain text report
• 1: Data could not be loaded
• 2: Invalid invocation

--data-file can be used to read a local OHLCV CSV instead of fetching data from the network; this parameter is mainly for testing.

Example Rules¶

The built-in engine provides an example rule: Bollinger Mean Reversion. The rule is at the daily level, entering when %B <= 0 and exiting when %B >= 1.

Use Cases and Notes¶

Suitable Use Cases¶

This plugin is suitable for the following scenarios:

• Using DSH web profile, wanting single-ticker research from chat or CLI
• Need to check the signal status of a rule on a specific ticker
• Need to perform daily-level backtesting on rule strategies
• Need to generate interactive, self-contained HTML reports
• Want to save researched strategies to the registry for reuse by id later

Unsuitable Scenarios¶

It is not suitable for actual order placement, position changes, or scenarios requiring a full trading system.

Usage Notes¶

• The plugin does not constitute investment advice; it only displays rule status and evidence; the decision is the user’s responsibility
• The plugin does not place orders or change positions
• The plugin runs with the permissions of the current DSH process; you should check the source code and license before installation
• Internet access is required to fetch Yahoo Finance data
• On the first run of the built-in Python engine, uv will sync python/.venv
• When --data-file reads a local CSV, it is mainly for testing or offline verification

Conclusion¶

The value of dsh-quant-workspace lies in placing several common steps of single-ticker quantitative research into one DSH plugin: daily data, rule backtesting, interactive reports, and strategy reuse. It maintains a read-only boundary and does not touch order placement or position changes, making it more suitable for research and review scenarios.

GitHub repository:

https://github.com/AllenCX/dsh-quant-workspace

Plugin directory page link:

https://www.skillhub.cn/plugins/AllenCX/dsh-quant-workspace