Getting Started
Start with the generic CLI path, then continue into a concrete worked example.
Open source planning software in Rust
Describe your problem → a running solver, today.
Tell your coding agent what has to be decided, what makes a plan invalid, and what makes it good. The solverforge-modeling skill turns that description into a SolverForge app with its domain model, hard and soft constraints, demo data, and an API, web, CLI, or MCP shell, and solves it before handing it back. The model stays in ordinary Rust you can read and change.
Apache-2.0, no account or licence key
Scalar and list planning variables
Eight worked planning apps
Write constraints like you write code
Readable rules, explicit tradeoffs, real application shapes.
Model shifts, routes, tasks, workers, vehicles, and inventories as ordinary application code. SolverForge keeps rule definitions, score analysis, and solver control close to the domain types your team already understands.
let required_skill = ConstraintFactory::<Plan, HardSoftDecimalScore>::new()
.shifts()
.filter(|shift: &Shift| shift.employee_idx.is_some())
.join((
Plan::employees_slice,
equal_bi(
|shift: &Shift| shift.employee_idx,
|employee: &Employee| Some(employee.index),
),
))
.filter(|shift: &Shift, employee: &Employee| {
!employee.skills.contains(&shift.required_skill)
})
.penalize(HardSoftDecimalScore::of_hard_scaled(1_000_000))
.named("Required skill");
Open source implementation
Solver core, app scaffolding, UI controls, and routing helpers you can inspect.
solverforge-cli
Scaffold a neutral project shell, then grow the domain with generators.
solverforge-ui
Ship scheduling views, retained-job controls, and embedded frontend assets.
solverforge-maps
Model matrices, route geometry, and map-backed planning systems.
Core Solver
Inspect the zero-erasure runtime, phases, moves, and score analysis tools.
Use cases
Eight planning surfaces you can watch, inspect, and run.
Machine scheduling, warehouse picking, crew rotas, asset readiness, hospital coverage, school timetabling, delivery routing, and field service routing. Each case carries its data shape, its constraints, a narrated walkthrough of the running app, and annotated screenshots of the same retained solve.
688 shifts / 50 employees / 28 days
Hospital
Coverage, qualifications, rest, availability, preferences, and fairness across dense workforce schedules.
300 lessons / 20 teachers / 10 rooms
Lessons
Two scalar decisions per lesson: which weekly timeslot fits the timetable, and which room fits the class.
82 deliveries / 10 vehicles / ordered routes
Deliveries
Route ownership, capacity, time windows, recommendations, and REST workflows on one inspectable surface.
48 visits / 6 technicians / road routes
Field service
Technician routes, skills, parts, shifts, territories, time windows, and road geometry in one retained workflow.
155 orders / 11 furnaces / 39 operators
Furnace
Furnace capacity, thermal sequencing, task operators, monitoring load, and shift rosters solved as one plan.
8 orders / 145 items / 5 trolleys
Order picking
Item assignment, bucket capacity, order splits, product locations, and closed walking routes on one warehouse canvas.
24 vessels / 56 days / 66 decisions
Fleet readiness
Maintenance, inspections, training, docks, technicians, parts, and daily readiness floors in one operating plan.
10 legs / 24 crew / 42 seats
Flight crew
Qualification, connectivity, availability, and rest and recovery rules solved as one rotation.
Evaluate SolverForge
Everything you need to decide is public.
- Licence The solver and the CLI are Apache-2.0: read the solver licence and the CLI licence.
-
Self-serve
cargo install solverforge-clifrom crates.io, then the agent skill from the CLI repository. No account or licence key. - Your problem shape Facts, entities, scalar and list planning variables, and hard and soft constraints you write yourself, delivered as an API, web, CLI, or MCP app. See Modeling & Generation and the maintenance walkthrough.
- Track record Eight worked use cases, the release notes for every line, and the full source on GitHub.