The Modular Monolith Paradigm

When to Employ This Paradigm

When you desire team autonomy similar to that of microservices, but without the operational overhead of a distributed system.
When release velocity is slowed by tangled dependencies between internal modules.
When a monolithic architecture is simpler to operate today, but there is a clear need to evolve toward a service-based model in the future.

Identify Modules: Define module boundaries that align with distinct business capabilities or Bounded Contexts from Domain-Driven Design.
Encapsulate Internals: Use language-level visibility modifiers (e.g., public/private), separate packages, or namespaces to hide the implementation details of each module.
Expose Public Contracts: Each module should expose its functionality through well-defined facades, APIs, or events. Forbid direct database table access or direct implementation calls between modules.
Enforce Architectural Fitness: Implement automated tests that fail the build if forbidden dependencies or package references are introduced between modules.
Plan for Evolution: Continuously track metrics such as change coupling and deployment scope to make informed decisions about if and when to split a module into a separate service.

An Architecture Decision Record (ADR) that maps module boundaries and defines the rules for any shared code.
Formal contract documentation (e.g., OpenAPI specs, event schemas) for every interaction point between modules.
Automated dependency checks and dedicated CI/CD jobs for each module to enforce boundaries.

Regression to a "Big Ball of Mud":
- Mitigation: Without strict enforcement, module boundaries will inevitably erode. Treat any boundary violation as a build-breaking error and maintain a disciplined approach to code reviews.
Shared Database Hotspots:
- Mitigation: High contention on a shared database can become a bottleneck. Introduce clear schema ownership, use view-based access to restrict data visibility, or implement data replication strategies to reduce coupling.

Skill not loading Check YAML frontmatter syntax and required fields

Token limits exceeded Use progressive disclosure - move details to modules

Modules not found Verify module paths in SKILL.md are correct