Introduction

In DSH (DeepSeek Harness) or agent workflows, stock research is often split into multiple steps: checking market data, reviewing value indicators, scanning for risks, examining institutional consensus, and performing valuation. If you manually stitch together APIs or ask for items one by one in the dialogue, the workflow tends to be fragmented.

dsh-invest-research is a plugin for DeepSeek Harness (DSH) designed to complete the “Market Snapshot → Value → Risk → Institutional Consensus → Valuation” five-layer research chain within a single call and output structured research report data.

It is maintained by eclipti, released under the MIT license, open source and free for personal research. The plugin is dedicated to quantitative research and data presentation and does not constitute investment advice.

What is this

dsh-invest-research is a full-chain investment research plugin for A-shares. It converges common research steps into a single command or model invocation, making it suitable for individual researchers who wish to complete single-stock research, multi-stock comparison, K-line technical analysis, sector heat queries, and Markdown report generation within DSH.

Core Features

The following introduces the verified five-layer research chain.

Market Snapshot

The Market Snapshot layer provides data such as current price, change percentage, PE, PB, market cap, turnover rate, and volume ratio.

Value Score

The Value Score layer uses the Buffett 7-dimensional framework to output a score out of 100.

Risk Scan

The Risk Scan layer outputs a risk score ranging from 0 to 100 and includes six types of alerts: ST, valuation, liquidity, capital flow, price, and volatility.

Institutional Consensus

The Institutional Consensus layer outputs rating distribution, number of institutions, average target price, highest target price, and lowest target price.

Conservative Valuation

The Conservative Valuation layer estimates based on DCF (Discounted Cash Flow) and provides a margin of safety. The default assumptions are 10% growth, 10% discount rate, and 3% perpetual growth.

Additional Capabilities

In addition to the five-layer chain, the plugin offers the following capabilities:

  • K-line + MA5/10/20 technical indicators
  • Hot sector rankings + main capital inflows
  • Multi-stock comparison table
  • Markdown report generation
  • Command line usage, no Harness required
  • Independent fault tolerance for each layer: failure of a single layer’s data source does not affect other layers
  • Supported scopes: A-share 6-digit codes, HK stock 5-digit codes, US stock codes

Installation and Enablement

Execute the following command in DSH to install the plugin:

dsh plugin --profile web add "github:eclipti/dsh-invest-research#main"

After installation, you can use the subsequent examples in DSH dialogue, shortcut commands, or command line entry.

The plugin’s declared runtime and dependency information is as follows:

engines.node >=18
peerDependencies：
@deepseek-ai/cordis >=4
@deepseek-ai/dsh-skill * (optional)
@deepseek-ai/dsh-tools *

Typical Usage

Model Invocation

You can directly raise a research request in the model dialogue, for example:

分析 600519

The model will automatically invoke invest_research to execute the five-layer full-chain research.

Shortcut Commands

You can also directly input shortcut commands in the input box:

/invest 600519

This method is used to directly obtain investment research briefs without model back-and-forth.

Command Line Invocation

The plugin supports command line usage without Harness. The following uses 600519 as an example.

First, run the complete five-layer JSON:

node cli.mjs 600519              # Complete 5-layer JSON

When you need to generate a Markdown report, use --report; the report will be output to the reports/ directory:

node cli.mjs 600519 --report     # Generate Markdown report to reports/

For multi-stock comparison, use --compare:

node cli.mjs --compare 600519 000858   # Multi-stock comparison table

When querying K-line and technical indicators, use --kline:

node cli.mjs --kline 600519 30   # K-line + Technical indicators

When querying hot sectors, use --sectors:

node cli.mjs --sectors 10        # Hot sectors

Applicable Scenarios and Notes

dsh-invest-research is suitable for personal quantitative research, public data aggregation, multi-stock comparison, and technical view quick checks. It is suitable for use as a research auxiliary plugin within a DSH workflow.

Please note before use:

  • The plugin only performs quantitative research and data presentation; it does not constitute investment advice.
  • The plugin does not provide buy/sell recommendations.
  • Scores are based on heuristic rules and are not a substitute for professional fundamental research.
  • Eastmoney (East Money) public APIs may be rate-limited or changed; single-layer fault tolerance is built-in, and you can extend backup data sources yourself.
  • The plugin will run with the current dsh process permissions; you should check the source code and license before installing.
  • The DSH plugin ecosystem includes independent sites such as the community directory, not an official app store.

Resources

GitHub repository:

https://github.com/eclipti/dsh-invest-research

If you need to search in the community directory, you can search by the plugin name dsh-invest-research.