The ecosystem philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” When setting up a local DSH instance, if you need to connect to A-share market data, perform quantitative stock screening, or run strategy backtesting, you can introduce dsh-stock-assistant. This plugin is maintained by fxxg023, registers 21 model tools, and provides agents with a complete set of capabilities ranging from market data queries to watchlist alerts.

Core Features

This plugin mainly solves three issues in A-share data analysis: data source integration, strategy backtesting, and real-time market monitoring.

  1. Data source coverage: It integrates three data sources.
    • Eastmoney (Dongfang Caifu): Provides market data, sector information, limit-up pool data, capital flow data, and news.
    • Tencent Quotes: Provides forward-adjusted K-line data.
    • akshare: Provides Dragon-Tiger List data and financial summaries.
  2. Model tools: It registers 21 tools, categorized as follows:
    • Market data and signals: stock_quote, stock_kline, stock_indicators, strategy.
    • Stock-profile recognition: stock_profile, including stock-profile recognition, strategy matching, and automatic backtesting recommendations.
    • Hotspot tracking: sector_rank, sector_flow, limit_up_pool, dragon_tiger.
    • Capital flow and financials: stock_flow, stock_finance, stock_news.
    • Quantitative functions: screen (multi-factor conditional stock screening), backtest (backtesting for 6 kinds of strategies).
    • Watchlist and monitoring: watchlist_add, watchlist_remove, watchlist_show, watch_start, watch_stop, watch_status, watch_alerts.
  3. Independent settings page: An independent “Stock Assistant” page appears in DSH settings, allowing users to edit watchlist stocks, backtesting/strategy parameters, and monitoring thresholds. Changes take effect after clicking “Save” in the bottom-right corner; “Initialize” can restore the default values.
  4. Monitoring alerts: Users can enable market monitoring. The system polls watchlist stocks every 30 seconds. When unusual price movements, limit-up/limit-down events, or volume surges occur, the events are recorded in the conversation log, and desktop bubble notifications are shown on Windows.

Installation and Prerequisites

Before installing, make sure your environment meets the following requirements:

  • Node.js ≥ 18 (used to build the client bundle).
  • Python 3 + the akshare library (used by the dragon_tiger and stock_finance tools; installation command is pip install akshare).
  • Windows: Desktop bubble notifications are available only on Windows; they are automatically skipped on other platforms.

Installation Steps

  1. Install the plugin by running the following command:
dsh plugin --profile web add github:fxxg023/dsh-stock-assistant
  1. After installation, locate the “Stock Assistant” section in DSH settings, then click Initialize or use the plugin directly.

Typical Usage

After the plugin is installed and initialized, you can invoke its tools during conversations. Below are several typical scenarios:

  1. Query market data:
    stock_quote 600000
    stock_kline 000001
  1. Stock-profile analysis and strategy matching:
    First, run stock_profile 600000. The plugin identifies the stock’s characteristics, recommends matching strategies, and provides backtesting results.
  2. Quantitative backtesting:
    backtest 600519 strategy ma_cross
  1. Watchlist monitoring:
    First add stocks through the settings page or watchlist_add, then start monitoring:
    watch_start

View the alert log:

    watch_alerts

Notes

  1. Data sources and disclaimer: The data used by the plugin are all obtained from unofficial public interfaces. The data are intended for learning and research only, and do not constitute any investment advice.
  2. Strategy parameter grouping: In the settings page, strategy parameters are displayed grouped by strategy name (for example, RSI oversold rebound). Remember to save after modifying parameters.
  3. Development-mode build: For local development and debugging, you must run pnpm install && pnpm build to generate lib/client.js; otherwise, the Web startup page will report a loading failure.

Closing

dsh-stock-assistant provides DSH users with a complete set of A-share quantitative analysis tools. It does not provide official data interfaces; instead, it relies on open-source projects and the aggregation capabilities of public interfaces. It is suitable for developers who want to build automated market analysis workflows in a locally and privately deployed environment.