Installation

Install the package

The milp-flare package is available on PyPI and can be installed with pip:

pip install milp-flare

Quickstart

This quickstart runs FLARE on a pair of formulations from the FormulationBench dataset (see Running FLARE on FormulationBench for more detail). It requires the following prerequisites:

• Docker installed and the flare-agent image built (see Docker)

• Claude Code authentication key on the host (see Agent Harnesses)

• FormulationBench Python package formulation-bench (see FormulationBench)

from pathlib import Path

from formulation_bench import Dataset
from milp_flare import FLARE, FormulationInput
from milp_flare.harness import ClaudeCodeHarness

ds = Dataset.load()
p1 = ds.problems[1]
a = p1.formulations["a"]
b = p1.formulations["b"]

harness = ClaudeCodeHarness(model="claude-opus-4-7", effort="medium")
flare = FLARE(harness=harness)

a_in = FormulationInput(
    formulation_md=a.render_markdown(), solve_py=a.gen_solve_py()
)
b_in = FormulationInput(
    formulation_md=b.render_markdown(), solve_py=b.gen_solve_py()
)

result = flare.verify(a_in, b_in, output_path=Path("runs/p1_a_b"))

See the User Guides for end-to-end tutorials.