AI Agent Hub
Back to plugins
🎮

agency-agents-zh-dsh-plugin

Fun & Dress-up Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install RHY006/agency-agents-zh-dsh-plugin

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install RHY006/agency-agents-zh-dsh-plugin in your terminal to install the expert role plugin from https://github.com/RHY006/agency-agents-zh-dsh-plugin, then restart dsh to activate all 268 registered skills.

About this plugin

The 268 Chinese AI expert roles spanning 20 departments were originally scattered Markdown files, forcing users to copy and paste each one into DeepSeek Harness for any real usage. This plugin bundles the entire library into a single dsh bundle, so one install registers every role as a native skill and eliminates the manual copy-paste workflow.

After installation the model auto-matches the right expert from conversation context, or you can trigger one explicitly with a slash command. Each skill carries the full persona, workflow, and deliverable text, with description used for routing and metadata preserving the Chinese name, color, and department. If you do not want all 268 roles loaded at once, the plugin supports filtering by department, an explicit whitelist, or keyword matching, and misconfigured values fail loud at startup rather than being silently ignored.

Ideal for Chinese-language AI developers building multi-agent workflows, product managers assembling domain-expert pipelines for a team, or anyone who wants to centralize a scattered prompt library into one manageable package.

Use Cases

  • Auto-match expert roles such as frontend dev, security, or marketing based on conversation context
  • Filter by department or whitelist to register only the expert subset your team needs
  • Trigger a specific expert workflow and deliverable spec via slash command

Best For

  • AI developers building Chinese multi-agent workflows
  • Product managers assembling domain-expert pipelines for teams
  • Engineers consolidating scattered prompt assets into one package