universalist
Algebra-Driven Design (ADD): model domains via minimal algebra + laws; encode in types; test laws.
When & Why to Use This Skill
This Claude skill implements Algebra-Driven Design (ADD) to help developers and architects build robust, mathematically sound software systems. By focusing on minimal algebraic structures and law-based verification, it enables the creation of type-safe domain models where illegal states are unrepresentable and system behavior is predictable and composable.
Use Cases
- Domain Modeling: Designing precise type systems for complex business logic to ensure data integrity and eliminate invalid states at compile-time.
- System Refactoring: Restructuring legacy codebases to align with algebraic boundaries, improving modularity through structures like Monoids, Functors, and Monads.
- Automated Law Testing: Implementing property-based and metamorphic testing suites to verify that implementation logic strictly adheres to defined mathematical laws.
- API Design: Creating highly composable and predictable software interfaces by deriving operations from core algebraic principles rather than ad-hoc branching logic.
| name | universalist |
|---|---|
| description | "Algebra-Driven Design (ADD): model domains via minimal algebra + laws; encode in types; test laws." |
Universalist
Intent
Use Algebra-Driven Design (ADD) to find the smallest algebra that fits the domain, encode it in types, and enforce laws with tests.
Core workflow (ADD)
- Frame the domain: observations, invariants, operations.
- Pick the minimal algebra; avoid overfitting.
- Define types: make illegal states unrepresentable.
- State laws: identity, associativity, distributivity, absorption, round-trip, homomorphism.
- Derive operations from the algebra (map/fold/compose) to reduce ad hoc branching.
- Refactor architecture to match algebra boundaries.
- Test the laws (property/model/metamorphic checks).
- Iterate: if laws are hard to state or test, the algebra is likely wrong.
Minimal-algebra decision guide
- Alternatives/variants → coproduct (tagged union)
- Independent fields → product (record/struct)
- Combine + identity → monoid
- Combine, no identity → semigroup
- Ordering/permissions → poset/lattice
- Add+multiply structure → semiring
- Structure-preserving map → functor
- Effectful apply → applicative
- Sequenced effects → monad
Testing expectations
- Include law/property tests for the chosen algebra.
- Prefer property-based checks when the repo already supports them.
- If not, write minimal deterministic law checks in the existing test framework.
- Don’t assume tooling exists; propose new dependencies only with user approval.
References
references/addd-overview.mdreferences/addd-sources.mdreferences/structures-and-laws.mdreferences/testing-playbook.mdreferences/case-studies.mdreferences/examples-haskell.mdreferences/examples-go.mdreferences/examples-typescript.md
Scripts
scripts/emit_law_test_stub.shprints illustrative law-test stubs for Haskell, Go, or TypeScript.