SpinForgeA Garmo Labs CompanyDiscuss a workload
UNISOLVE / DEVELOPER GUIDE

One entry point.
Three ways to work.

Current source: UNISOLVE 0.16.1, source G. The supplied inventory reports 82 registered profiles, 65 distinct entry functions, 133 public Python exports, 38 CLI operations and nine stdio MCP tools. Obtain the package and its release manifest separately.

Current source G: UNISOLVE 0.16.1, 82 profiles, 65 entry functions, 133 Python exports, 38 CLI operations and nine stdio MCP tools. All profiles were exercised through the installed-package campaign; optional declines and incomplete outcomes remain explicit. Current evidence ↗
FROM THE SUPPLIED DISTRIBUTION

Install once.
Call the supported task.

Python 3.10 or later. The wheel/source package must be obtained separately; this website contains marketing content and local command preparation, not the solver.

LOCAL EXAMPLE / INSTALLED PACKAGE
# After installing the agreed UNISOLVE package:
unisolve doctor
unisolve formats
unisolve solvers
unisolve solve-any instance.cnf --time 30 --cert result.json
unisolve verify result.json --require-optimal

A verify command with --require-optimal rejects incomplete proof outcomes. That does not mean the preceding budgeted solve is guaranteed to produce a complete result.

LOCAL COMMAND BUILDER

One input.
Checked acceptance.

Prepare the documented solve-any command and matching solve_any MCP arguments. No file is read, uploaded or executed by this page.

File type helps fill the example filename. solve-any performs actual detection in the installed package. The supplied MCP example does not document a time-budget argument, so the generated MCP JSON does not invent one.

LOCAL EXAMPLE / INSTALLED PACKAGE
Select the input and prepare your commands.

Generated syntax only. The stdio MCP tool is solve_any. Its exposed operations and permissions are configured in your installed environment.

DIRECT USE

Terminal. Python.
Stdio MCP.

All work without a display, browser, TTY or interactive prompts.

LOCAL EXAMPLE / INSTALLED PACKAGE
unisolve solve-any problem.opb --require-optimal --cert answer.json
unisolve solve-any problem.qdimacs --strategy race --time 60 --cert answer.json
unisolve router solve problem.unknown --time 60 --cert answer.json
unisolve mcp --workspace /path/to/instances

MCP tool: solve_any. Example arguments: {"file":"/path/to/problem.cnf","strategy":"race","require_optimal":true}. This is a stdio server, not an invented hosted URL.

PYTHON INTERFACE

An exact value.
An explicit check.

This small functional example is from the supplied update. It demonstrates an admitted irrational result, not general nonlinear completeness.

LOCAL EXAMPLE / INSTALLED PACKAGE
from unisolve import parse_text, solve, SolveConfig, create_certificate, verify_certificate
import json

source = {
    "kind": "polynomial", "sense": "min",
    "variables": [{"name": "x", "kind": "real", "lb": 0, "ub": 2}],
    "objective": [{"powers": [1], "coefficient": 1}],
    "rows": [{"polynomial": [{"powers": [2], "coefficient": 1}],
              "relation": "=", "rhs": 2}],
}
problem = parse_text(json.dumps(source), "json")
config = SolveConfig(mode="exact", backend="cpu", time_limit=5)
result = solve(problem, config)
assert result.status == "optimal" and result.objective is None
assert result.objective_algebraic["kind"] == "real-algebraic"
assert verify_certificate(
    create_certificate(problem, result, config), require_optimal=True
)["valid"]

The required ordinary objective is null for the irrational value; its exact interpretation is retained separately. Execution depends on the obtained package and admitted source scope.

PACKAGE IDENTITY

Use the agreed release.
Keep the evidence separate.

The current account identifies UNISOLVE 0.16.1 source G. Use its supplied distribution and manifest. The commands here prepare documented operations; the website does not contain or install the solver. Environment and dependency limits remain release-specific.

Current candidate and historical tests
YOUR MODEL. A DEFINED RESULT.SPINFORGE / A GARMO LABS COMPANY

Make the
next move.

Start with the model, the environment
and the evidence your workflow needs.

patrick@garmolabs.com

FIND YOUR CAPABILITY