Introduction

After DSH provides Agent kernel and Session capabilities, a concrete issue emerges: while Agents can call tools, A-share research on market panorama, watchlist grouping, individual stock K-lines, valuation, expert conversations, and research reports still requires a clear interface and data flow to support them.

hanai-investment-dsh is a local-first A-share research workbench powered by the DeepSeek Harness (DSH) Agent kernel. It integrates market viewing, watchlist creation, individual stock analysis, open conversations, research initiation, and continuous verification into a single workflow. Below, we introduce its positioning, verified capabilities, environment requirements, and usage boundaries.

What This Is

hanai-investment-dsh is a DSH plugin positioned as a local-first A-share research workbench.

  • License: MIT.
  • Maintainer Information: Hanai Worth contributors.
  • Environment Requirements: Node.js ^22.19.0 || >=24.0.0, with packageManager set to pnpm@11.7.0.
  • Repository URL: https://github.com/hancao97/hanai-investment-dsh.

It is designed for users who need to integrate A-share market data, watchlists, valuations, K-line observations, expert conversations, and research reports into a single local workflow. As indicated by its positioning, it is not merely a market display tool but organizes functions around “viewing the market, building watchlists, analyzing individual stocks, discussing, researching, and continuing verification.”

Core Features

Market Panorama

The market panorama includes:

  • Six major indices
  • Limit up/down and number of rising/falling stocks
  • Trading volume for both markets
  • Industry/concept treemap
  • Four types of lists

This section breaks down the day’s market structure into indices, breadth, trading volume, and sector heat maps, making it easy to view the overall picture before diving into individual stocks.

Watchlist Grouping

Watchlist grouping supports:

  • Creating, renaming, and deleting custom groups
  • Adding, moving, and removing stocks
  • Three-state sorting

In practice, you can search for stocks first, then add, move, and maintain them by group. After refreshing the current group, you can compare market conditions, valuations, and performance since addition.

Individual Stock Market Data and K-lines

The individual stock section supports:

  • Real-time, daily K, weekly K, and monthly K charts
  • Forward-adjusted prices
  • Dual moving average mode
  • Turning point markers and historical evidence

Note that turning points here are volume-price observation prompts and do not equate to buy/sell points or revenue promises.

Reasonable Valuation and Value Curve

Value-related displays on the individual stock page include:

  • Valuation summary
  • Five-dimensional radar chart
  • Independent display of price/master value curves

Missing fields are shown as — or hidden, and null values are not interpreted as 0.

Expert Open Conversations

Expert open conversations do not require selecting a stock first. You can directly choose an expert to have ongoing discussions about industry cycles, business models, recent events, or decision-making issues.

Each conversation is bound to an independent expert snapshot and DSH Session.

The five conversation experts are:

  • Duan Yongping
  • Hun Jianglong
  • Charlie Munger
  • Warren Buffett
  • Sun Yucheng Perspective

The Sun Yucheng Perspective is an AI simulation boundary. A boundary notice is provided when creating a conversation, with identity disclosure completed in the first response. Specific time-sensitive facts must be verified through retrieval first.

Master Research

The master research process involves selecting a stock and a master, with the Agent retrieving and verifying public information to generate an archivable Markdown report.

Regarding continuous verification:

  • Open conversations and research follow-ups reuse the DSH Session.
  • Research can view tool processes.
  • Research allows explicit report revisions.

Full Market Backtest Research

The documentation mentions that FULL_A_TURNING_V0 research has frozen 16 rules and analyzed 9,813,872 valid tradable daily K-bars.

Note the following:

  • The strict launch threshold for this round is 0/16.
  • The product has not added any newly verified high win-rate buy points.
  • Full market research data is up to 2026-08-20 and can only be used as retrospective evidence.
  • The history of delistings on the Beijing Stock Exchange, daily ST status, and complete delisting returns are still incomplete.
  • Screenshot percentages are historical conditional frequencies up to 2026-08-20 and do not constitute investment advice.

Installation and Enabling

The verified documentation does not provide official installation commands, so this article does not specify concrete commands or construct commands like dsh plugin add. The specific installation method should be based on the repository instructions or DSH plugin directory commands.

Before installation, it is recommended to confirm the environment:

Node.js: ^22.19.0 || >=24.0.0
packageManager: pnpm@11.7.0

The license is MIT. Since the plugin runs with the current DSH process permissions, you should check the source code and license before installation, and confirm that its tool calls and data processing align with your security boundaries.

Typical Usage

Below is the usage path based on the documentation steps.

  1. View the Market

Check the six major indices, market breadth, industry/concept heat maps, and lists for gainers, losers, trading volume, and turnover rates.

  1. Build a Watchlist

Search for stocks, add, move, and maintain them by group; refresh the current group and compare market conditions, valuations, and performance since addition.

  1. Analyze Individual Stocks

Combine forward-adjusted K-lines, moving averages, volume-price observation markers, market snapshots, fundamentals, reasonable valuations, and value curves to build context.

  1. Open Conversations

Without selecting a stock, directly choose an expert to have ongoing discussions about industry cycles, business models, recent events, or decision-making issues.

  1. Initiate Research

Select a stock and a master, with the Agent retrieving and verifying public information to form a complete, archivable Markdown report.

  1. Continue Verification

Open conversations and research follow-ups reuse their respective original DSH Sessions; research can view tool processes or explicitly revise reports.

Applicable Scenarios and Notes

Suitable for:

  • Developers who want to add an A-share research interface within DSH.
  • Researchers who need to integrate watchlists, individual stocks, valuations, and research reports into a single local workflow.
  • Users who wish to bind expert conversations and research follow-ups to independent Sessions.

Not suitable for:

  • Individuals expecting the plugin to directly provide buy/sell points or revenue promises.
  • Those who interpret historical backtest ratios as future probabilities or investment advice.

Before use, several boundaries should be kept in mind:

  • Turning points are volume-price observation prompts and do not equate to buy/sell points or revenue promises.
  • Full market backtest data is up to 2026-08-20 and can only be used as retrospective evidence.
  • The strict launch threshold for FULL_A_TURNING_V0 is 0/16 in this round, so it should not be understood as newly verified high win-rate buy points.
  • The Sun Yucheng Perspective is an AI simulation boundary; specific time-sensitive facts must be verified through retrieval first.
  • Missing fields are shown as — or hidden, and null values are not interpreted as 0.
  • The plugin runs with the current DSH process permissions; check the source code, license, and tool call permissions before installation.

Conclusion

The value of hanai-investment-dsh lies in applying DSH’s Agent capabilities to concrete A-share research objects: market, watchlists, individual stocks, conversations, research, and continuous verification can all be integrated within a single workflow. It emphasizes evidence, context, and archivable reports rather than directly providing trading conclusions.

Repository URL: https://github.com/hancao97/hanai-investment-dsh.