hello-dsh

Guides first-time users through DeepSeek Harness plugin architecture and lifecycle concepts.

96|8|Updated Aug 13, 2026
One-click install
npx skills add https://github.com/pingfanfan/hello-dsh --skill hello-dsh-pingfanfan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: hello-dsh
Source: https://github.com/pingfanfan/hello-dsh/tree/main/examples/skills/hello-dsh
Command: npx skills add https://github.com/pingfanfan/hello-dsh --skill hello-dsh-pingfanfan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? New DeepSeek Harness users lack a structured entry point to understand its plugin-based architecture, verify that the skill system works, and learn how to extend the platform with Markdown skills or TypeScript code plugins. ## Core Features & Use Cases - Layered Onboarding: Explains DSH's "Everything is a Plugin" design in four progressive layers, from architecture overview to the Cordis spatiotemporal composability theory behind it. - Live Lifecycle Verification: Walks users through deleting and restoring the skill file to observe hot-reload behavior without restarting the process. - Code Plugin Pitfall Guide: Documents three verified beginner errors when writing TypeScript plugins, including the defineTool output schema requirement, the required-field restriction, and the export default inject bug. - Use Case: A developer types "hello dsh" after installing DeepSeek Harness and receives a guided, step-by-step introduction that confirms the skill pipeline works and teaches them how to build their first plugin. ## Quick Start Say "hello dsh" to start the guided first lesson on the DeepSeek Harness plugin system.

Frequently Asked Questions about hello-dsh

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get started with DeepSeek Harness plugins?▼

Start by saying "hello dsh" to trigger the guided first lesson, which walks through the plugin architecture layer by layer. You can then choose between writing a Markdown skill in about five minutes or a TypeScript code plugin for deeper extensions.

How do I verify the DSH skill system is working?▼

Delete the skill directory and ask the model about it, then restore the file and ask again. The skill appears and disappears without any restart, confirming the filesystem skill plugin and hot-reload lifecycle are functioning.

Why are DSH skills not working in the web UI?▼

The web UI disables skills by default while headless mode enables them. Launch with npx @deepseek-ai/dsh web --patch ./enable-skills-in-web.yml to explicitly turn on skill support, verified on version 0.1.0-rc.6.

What are common errors when writing a DSH code plugin?▼

Three verified pitfalls: tools must be wrapped in defineTool with an output schema and render declaration, the required field only accepts true, and modules must not use export default or the loader drops inject. Parameters use flat field mapping while output.schema is standard JSON Schema.

What is the difference between a DSH skill and a code plugin?▼

A skill is a Markdown file read by the skill-filesystem plugin, suited for changing judgment criteria, output formats, and workflows in minutes. A code plugin is a TypeScript module for registering new tools, connecting external services, and hooking lifecycle events.