Introduction

In the DSH workflow, mathematical calculations often require more than one backend: simple expressions and symbolic simplification can be handled by mathjs; integrals, equation solving, limits, series, and special functions might require Wolfram Mathematica; when manifolds, coordinates, tensors, Christoffel/Riemann/Ricci are involved, SymPy.diffgeom is more appropriate. If you search for and install plugins one by one, the configuration cost will increase.

Below, I introduce dsh-math-suite. It is a DeepSeek Harness (DSH) plugin aggregation package designed to install three math plugins at once.

What is it

dsh-math-suite is maintained by hatter123 with an MIT license.

It aggregates the installation of the following three plugins:

  • dsh-math
  • dsh-wolfram
  • dsh-diffgeom

After the installation above, three types of tool entries will appear in DSH: math, wolfram, and diffgeom. Each tool corresponds to a different backend, and therefore has different prerequisites.

Core Features

math：mathjs

The dsh-math plugin provides the math tool with the mathjs backend, supporting:

  • Numerical evaluation
  • Simplification
  • Symbolic differentiation

This plugin has no prerequisites and is a pure JS implementation.

wolfram：Wolfram Mathematica

The dsh-wolfram plugin provides the wolfram tool with the Wolfram Mathematica backend, supporting:

  • Integrals
  • Equation solving
  • Limits
  • Series
  • Special functions

This tool requires an external environment providing the Wolfram Engine, or a Mathematica that provides wolframscript.

diffgeom：SymPy.diffgeom

The dsh-diffgeom plugin provides the diffgeom tool with the SymPy.diffgeom backend, supporting manifold, coordinate, and tensor mathematics, including:

  • Exterior derivative d
  • Wedge product ∧
  • Lie derivative
  • Covariant derivative
  • Christoffel/Riemann/Ricci

This tool requires Python 3 and SymPy.

Installation and Usage

First, confirm that the current DSH environment meets the peer dependency requirements of this package. The version range declared in package.json is:

"@deepseek-ai/dsh-tools": ">=0.0.1-rc.1 <0.1.0 || >=0.1.0-rc.1 <0.2.0-0"

Then, execute the installation command:

dsh plugin --profile web add dsh-math-suite

This command installs dsh-math-suite into the web profile. After installation, use math, wolfram, and diffgeom according to the tool entry points provided by DSH.

Prerequisites

Check dependencies by tool to avoid mistakenly assuming all mathematical capabilities are available.

  1. dsh-math: No prerequisites.
  2. dsh-wolfram: Requires Wolfram Engine, or a Mathematica providing wolframscript.
  3. dsh-diffgeom: Requires Python 3 and SymPy. To install SymPy, you can use:
python -m pip install sympy

When prerequisites are missing, graceful degradation occurs by tool: dsh-math is always available; dsh-wolfram and dsh-diffgeom return Error: information when the corresponding backend is missing.

Applicable Scenarios and Notes

This plugin package is suitable for developers who need to put different math backends into the same DSH workflow. It solves the installation and entry aggregation problem: install once, and three tools enter the profile together.

Usage notes:

  • Plugins run with the current dsh process permissions; check the source code and license before installing.
  • dsh-wolfram depends on an external Wolfram environment.
  • dsh-diffgeom depends on Python 3 and SymPy environments.
  • License is MIT.

Conclusion

The value of dsh-math-suite lies in aggregating three types of math tools—math, wolfram, and diffgeom—into a single DSH plugin package, while retaining the features of independent availability and graceful degradation for each backend.

GitHub Address:

https://github.com/hatter123/dsh-math-suite