When debugging a DeepSeek Harness Agent or troubleshooting local environment issues, the Agent needs to know the specific configuration of the current runtime environment. Manually querying system information is inefficient and error-prone. The dsh-tool-sysinfo plugin provides DSH with a read-only local environment snapshot capability, allowing the Agent to directly access core system data.

Plugin Overview

This is a system information tool plugin for DeepSeek Harness, maintained by developer shiyi-0x7f. It provides the Agent with a tool named system_info, which can read critical information such as the operating system, CPU, and memory.

  • License: MIT
  • Dependencies: Zero external dependencies, zero network requests, zero write operations

Core Features

The tool provides a read-only snapshot of the local machine environment, including the following information:

  • Read system information (OS/architecture)
  • Read the hostname
  • Read uptime
  • Read CPU model and core count
  • Read memory usage
  • Read the Node version
  • Read non-loopback IPv4 addresses

Installation and Activation

Use the official install command to add the plugin:

dsh plugin --profile web add github:shiyi-0x7f/dsh-tool-sysinfo

After installation, restart dsh --profile web (or dsh-shell). In the chat interface, ask the Agent to call the system_info tool to see the result.

Note: There is a known junction bug on Windows with pnpm 10.x. It is recommended to use pnpm ≥ 11 or install directly via the GitHub link.

Implementation Details

This plugin is a single-file, zero-build plugin. All data comes from the Node.js native modules node:os and process. The plugin declares the composition layer (dsh.bundle) via cordis.patch.yml, and after installation is automatically added to the profile layer stack.

Ecosystem Context

The core design philosophy of DeepSeek Harness is “everything is a plugin”. The community directory (SkillHub) is a third-party-maintained plugin aggregation site and has no official affiliation with DeepSeek or Huanfang AI. Before using community plugins, it is recommended to review the source code and license.