Summary
An optimisation model can live in the same conversation as the code that generates it.
One call takes the problem and returns the solution, so an LP or MIP model becomes an argument you assemble where the data already is. The quadratic path carries the constraints the tool states outright: QP accepts continuous variables only, so there is no MIQP, and values in Q must be doubled to account for the 0.5 factor in the objective.
What it is
A single-tool front end to the HiGHS solver: it takes a problem definition and an options object and returns the solve.
What you get
- Linear programming (LP) and mixed-integer programming (MIP) problems solved by HiGHS from a problem you pass in
- Convex quadratic objectives of the form minimize c^T x + 0.5 x^T Q x, with Q positive semidefinite
- Solver behaviour set per call through an options object carried alongside the problem
Requirements
A model you already hold as data — variables, constraints and objective — and nothing else: no account, no key, and no solver to install alongside it, since HiGHS is what the tool calls.
Setup effort
One command — npx highs-mcp