AI Agent Hub
Back to plugins
🖥️

dsh-dshx

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install YZz-S/dsh-dshx

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

Run dsh plugin install YZz-S/dsh-dshx in your terminal to install; source at https://github.com/YZz-S/dsh-dshx .

About this plugin

DSH Web GUI is a lightweight agent frontend, but once you want to connect external MCP servers, maintain skill bundles, define custom slash commands, or curate context memory, there is no unified surface to operate and monitor those resources. dshx fills that gap by packaging an MCP client hub, skill management, slash-command registry, memory store, and prompt library into a single mountable extension suite.

The MCP hub is the centerpiece: it speaks stdio, Streamable HTTP, and SSE out of the box, with a per-server state machine, exponential-backoff auto-reconnect, a ring-buffer log, and a live tool-discovery panel. Discovered tools are bridged to the model as native tools named mcp____. The web management console rides on the same local DSH web server (port 3080), offering five tabs—one per module—plus a JSON API under /dshext/api/* for scripted workflows. Every module also exposes model-facing tools (memory_write, skill_create, command_define, and so on), so you can simply ask the agent in natural language to create a skill, write a memory note, or define a command for you.

dshx is well suited for developers who already run the DSH Web GUI and want to extend their agent declaratively—particularly when juggling multiple MCP servers, sharing memory and skills across projects, or codifying recurring workflows as named slash commands. It is pure ESM JavaScript with no build step and zero additional npm dependencies.

Use Cases

  • Connect and manage multiple MCP servers from one DSH Web GUI console
  • Share memory, skills, and custom slash commands across projects
  • Review and edit prompt sections and persona via a local web dashboard
  • Ask the agent in natural language to create skills, write memories, or define commands

Best For

  • Developers extending the DSH Web GUI with declarative agent capabilities
  • Engineers who frequently switch MCP servers or maintain multi-tool pipelines
  • Users who want to codify recurring workflows as slash commands or skill bundles