AI Agent Hub
Back to plugins
oh-my-dsh preview

oh-my-dsh

Model Inference Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install amplifthq/oh-my-dsh

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

Install it in DeepSeek Harness by running dsh plugin install amplifthq/oh-my-dsh from its open-source repository at https://github.com/amplifthq/oh-my-dsh.

About this plugin

DeepSeek Harness gives you a strong plugin foundation, but its default setup can still feel like a newly handed-over empty room: the capabilities are there, while the everyday coding workflow is not fully configured yet. oh-my-dsh addresses that gap. It is not a separate fork; it is a curated distribution built on top of upstream, pre-arranging the profiles, boundaries, and extension flows that make the harness feel practical from the very first session.

Its strength is not just bundling more plugins, but adding a layer of control to agentic coding: language-server navigation, previewable semantic rename, safer web fetch by default, approval-gated skill distillation, and lazy reuse of existing MCP servers, instructions, commands, and hooks. The model can propose activation instead of demanding full access, which keeps extension available while reducing the blast radius.

This plugin is a good fit if you want a more opinionated starting point on top of DeepSeek Harness, or if you already use agent setups around tools such as Claude, Cursor, or Codex and would rather reuse them than rebuild. It especially suits developers who care about permission boundaries, recoverable edits, and long-term maintainability.

Screenshots

Use Cases

  • Quickly turn DeepSeek Harness into a more practical coding environment.
  • Reuse existing MCP servers, instructions, hooks, and commands without rebuilding your setup.
  • Get more controlled renaming, web fetching, and plugin activation workflows.

Best For

  • Developers who rely heavily on coding agents
  • Users with existing Claude, Cursor, or Codex configurations
  • Teams that care about permission boundaries and recoverable actions