Preamble

The core philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” When building an enterprise-grade AI resource management system, unified management is required for IAM, MCP services, model gateways, and usage metering. This plugin is maintained by 01men and aims to provide a “Rongqi Custom Track (Gate 01)” — a DSH-based collaboration frontend for AI agents and humans. It integrates the cordis plugin form into DSH with one click, serving as a unified entry point for management console and business data interactions.

Plugin Positioning

  • Name: Rongqi Custom Track (Gate 01)
  • Type: Workflow / Enterprise Resource Management Platform
  • License: MIT
  • Core Value: Unifies business domains in the DSH ecosystem (IAM, model gateways, skills marketplace, etc.) into a single collaboration frontend, enabling coordinated operations between agents and humans.

Core Features

This plugin builds a complete Industry AI Workbench (IAW), including the following capabilities:

  • Industry AI Workbench (IAW): Provides a five-space information architecture and industry graph packages.
  • Unified Authentication and IAM: Supports organizational accounts and unified authentication.
  • DingTalk Device Flow Login: Supports DingTalk device flow login as a connector.
  • MCP Deployment Service: Provides deployment, monitoring, and governance for MCP services.
  • Skills and Plugins Marketplace: Provides management and integration for the Skill/plugin marketplace.
  • Model Access Gateway: Unified access to large language models with transparent usage metering.
  • Business Data Layer: Provides business data support based on scene graphs.
  • Collaboration Frontend Components: Includes plugin-panel-core (panel), plugin-rq-card (execution card), and plugin-dingtalk-bridge (DingTalk bridge).

Installation and Enablement

Use the following command to add the plugin to DSH with one click:

dsh plugin add github:01men/RQ-DSH / npm @ybkk/gate-01

After installation, start the service to access the console.

Typical Usage

After starting the platform, the default URL is http://127.0.0.1:7300.

  1. Basic Startup:
    npm start

On first startup, baseline initialization is performed and a platform administrator account is generated.
  1. Demo Mode:
    If you need to quickly evaluate or demonstrate the platform, start it with seed data:
    DEMO_SEED=1 npm start

In demo mode, complete demo data and accounts are automatically generated, and the unified password is `Ybk@2026`.

  1. Management Console:
    Open http://127.0.0.1:7300 in a browser to enter the management console.

  2. Feature Self-Test:

    npm run selftest

Architecture and Collaboration Rules

  • Plugin Form: The plugin runs in cordis form. After connecting to the host, all data is accessed through a local proxy (zero browser CORS).
  • Collaboration Iron Rule: Plugins are prohibited from directly accessing each other’s data. State changes must trigger events through the event bus. For example, the iam.user.frozen event causes authn to revoke tokens.

Notes

  • M0 Compliance Convergence: The platform has deprecated usage-based billing, resale, and revenue-sharing features. The business model has been adjusted to team-based plus domain asset pricing.
  • Wallet Subsystem: The wallet/revenue-sharing subsystem has been taken offline and archived.
  • Environment Variables: The organizational account password is taken from the ADMIN_PASSWORD environment variable. If not set, a random password is generated and written to the logs.
  • Offline Support: Single-file and self-contained, with zero external CSS/JS/font dependencies, and supports offline use in intranet or private deployment environments.
  • Intranet Deployment: To increase GitHub token limits or use private mirrors in intranet environments, configure them via environment variables.

Summary

01men/rq-dsh is a complete enterprise-grade AI resource management solution built on DSH. Through its plugin architecture, it integrates core capabilities such as identity authentication, model access, and skills marketplace into a unified collaboration frontend, making it suitable for teams that need to build internal AI governance systems or deploy private AI platforms.