Introduction

DeepSeek Harness (DSH) uses a plugin-based architecture designed to decouple general capabilities from domain-specific logic. When an agent needs to handle business logic—such as pricing strategy, profitability calculation, or commercial handoff—general model reasoning alone is often insufficient to provide structured decision support. The dsh-business plugin fills this gap by providing a set of evidence-based business strategy and commercialization tools for the model reasoning layer.

Plugin Positioning

dsh-business is positioned as a cross-dimensional “Business Strategy Layer.” It is not responsible for specific need discovery, product delivery, or CRM operations. Instead, it spans four stages—requirements, product, marketing, and commercialization—and translates customer value, product capabilities, target segments, and operational outcomes into interpretable business models, pricing structures, and profitability paths.

Installation and Dependencies

Before installing this plugin, ensure the environment satisfies the following dependency requirements:
* DeepSeek Harness version: 0.1.5-rc.2
* Cordis version: 4.0.2
* Peer dependencies: @deepseek-ai/dsh-fs and @deepseek-ai/dsh-tools (version 0.1.5-rc.2)

When installing, the correct profile must be specified (for example, web), because the plugin is not enabled automatically. The installation command is:

dsh plugin --profile web add github:winyh/dsh-business

After installation, restart the corresponding profile process.

Core Features

This plugin provides a series of decision-support tools for business decisions, covering the full process from strategy formulation to implementation review:

  1. Business Strategy and Commercialization: Provides an overall strategy framework.
  2. Business Model Analysis: Decomposes and evaluates business models.
  3. Pricing Architecture Design: Designs price tiers and combinations.
  4. Channel Economics Evaluation: Analyzes the costs and benefits of each channel.
  5. Profitability Analysis: Calculates contribution margins and profitability assumptions.
  6. Elevator Pitch Generation: Quickly generates business pitch copy.
  7. Business Plan Creation: Produces a structured business plan.
  8. Business Handoff Management: Manages business handoffs across plugins or teams.
  9. Business Closed-Loop Review: Checks whether feedback from discovery, product, business, and growth forms an actionable closed loop.

Workflow and Typical Usage

When building a cross-plugin collaboration chain, the following tools can be used for linking and validation:

  • Receive Handoff: When starting a cross-plugin workflow, use business_handoff_receive to receive business handoff data.
  • Scan Structured Artifacts: Use business_artifact_index to scan structured artifacts in the project.
  • Generate Handoff Documentation: Use business_commercial_handoff to convert pricing reviews and profitability reviews into versioned business handoff documents.
  • Full Validation: For maintainers, run pnpm run verify to execute full type checking, unit tests, build validation, and runtime plugin validation.

Applicable Scenarios and Notes

This plugin is suitable for scenarios that require embedding business logic into agent workflows, such as when generating a business plan, evaluating pricing strategies, or reviewing business closed loops.

Notes:
* Permissions and Profiles: It must be installed in the correct profile (such as web) to be enabled. Installing it to default does not make it effective in web mode.
* No Execution Responsibility: The plugin does not handle requirement discovery, product delivery, marketing content execution, or CRM operations.
* Decision Support, Not Approval: The provided tools are only used to provide business judgment and rules. They do not automatically approve prices, discounts, or revenue commitments.
* Private File Exclusion: Private docs/ files are excluded from Git and package contents.

Ecosystem and Resources

This plugin is part of the DeepSeek Harness ecosystem and is designed to transform real business decision processes into traceable evidence chains. More details and source code can be found in the GitHub repository.