AI Agent Hub
Back to plugins
🖥️

dsh-plug-mcp

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Casually/dsh-plug-mcp

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

Run dsh plugin install Casually/dsh-plug-mcp in your terminal to install the plugin from https://github.com/Casually/dsh-plug-mcp, then restart DSH to find the MCP panel under Settings > Plugins.

About this plugin

The MCP ecosystem is growing fast, yet wiring up each service still means hunting for repos, reading docs, hand-editing YAML, and restarting the gateway -- a loop that is both tedious and fragile. dsh-plug-mcp pulls the entire workflow into the DeepSeek Harness web UI, so everything happens right in your browser.

Open Settings > Plugins > MCP to search GitHub by topic, sort by stars or recency, and inspect a detail panel that surfaces the license, rendered README, and auto-detected launch commands (npx, uvx, docker run). Pick a candidate, confirm, and the service is installed via a text-level append to the home-layer patch file, complete with automatic backup and read-back validation. Hot-reload kicks in on save, so a restart is rarely needed. The installed list lets you remove a service with one click, and a built-in proxy panel handles HTTP, SOCKS, or custom proxy setups behind restricted networks with an instant connectivity test.

If you are building multi-agent workflows on DeepSeek Harness and keep adding or swapping MCP tool servers, this plugin turns what used to be a YAML-surgery session into a few browser clicks -- no shell, no guesswork, no broken configs.

Use Cases

  • Quickly discover and onboard new MCP tool servers from GitHub without manually reading docs or editing YAML.
  • Frequently add or remove multiple MCP services in one gateway and want a visual list to trace and delete them with one click.
  • Working behind a firewalled network that requires an HTTP or SOCKS proxy before GitHub API searches and installations succeed.

Best For

  • Developers or platform admins building multi-agent workflows on DeepSeek Harness.
  • Engineers who swap MCP tool servers often and want to avoid endless YAML edits and startup troubleshooting.
  • DevOps leads who want to centralize and visualize the team's MCP service configuration to cut operational overhead.