agentleFS
MesteriisMesteriis/Engineering-Bible-AI/skills/architecture-principles/SKILL.md

architecture-principles

Skill0 starsChanged 4 months agoMITView on GitHub

Guides module boundaries, dependency direction, and domain structure. Do not use for local edits that leave architecture unchanged.

We split this skill into the rules it gives an agent

16 rules. 0 are common, 16 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 16

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not force Clean Architecture, DDD, CQRS, Event Sourcing, Hexagonal Architecture, microservices, or any other style unless:

    line 18
  2. Distinctive:

    Prefer dependencies pointing toward stable business rules.

    line 28
  3. Distinctive:

    When boundaries are already messy, make the change without making them worse.

    line 60
  4. Distinctive:

    Use DDD where domain complexity justifies it.

    line 64
  5. Distinctive:

    Prefer DDD concepts when the codebase has:

    line 66
  6. Distinctive:

    Do not use DDD ceremony for simple data plumbing.

    line 75
  7. Distinctive:

    Use domain language in names.

    line 79
  8. Distinctive:

    Keep invariants close to the domain model.

    line 80
  9. Distinctive:

    Keep application orchestration separate from domain decisions.

    line 82
  10. Distinctive:

    Keep infrastructure adapters outside the domain model.

    line 83
  11. Distinctive:

    Keep bounded contexts explicit.

    line 84
  12. Distinctive:

    Do not leak persistence concerns into domain behavior.

    line 85
  13. Distinctive:

    Use cases should express application behavior.

    line 90
  14. Distinctive:

    Do not introduce interfaces for one implementation unless the boundary is meaningful.

    line 125
  15. Distinctive:

    Before finishing an architectural change, verify:

    line 141
  16. Distinctive:

    no layer gained unrelated responsibilities.

    line 149

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Skill: architecture-principles

Purpose

Design and modify software while preserving boundaries, dependency direction, domain clarity, and long-term maintainability.

This skill covers Clean Architecture, Hexagonal Architecture, DDD, modularity, layering, dependency inversion, and boundary discipline.

First Rule

Respect the existing architecture.

Do not force Clean Architecture, DDD, CQRS, Event Sourcing, Hexagonal Architecture, microservices, or any other style unless:

  • the repository already follows it;
  • the change clearly requires it;
  • or the user explicitly requests it.

Architecture is a tool. It is not a costume party for simple CRUD.

Dependency Direction

Prefer dependencies pointing toward stable business rules.

General direction:

Transport / UI / CLI
  -> Application / Use Cases
      -> Domain
Infrastructure / Persistence / External APIs
  -> Application or Domain abstractions

Rules:

  • Domain rules must not depend on transport details.
  • Domain rules must not depend on database models unless the repository intentionally uses active record style.
  • Application/use-case code coordinates behavior but should not become a dumping ground.
  • Infrastructure implements details and adapters.
  • Transport layers should be thin.

Boundary Rules

Do not mix:

  • routing and business policy;
  • validation and persistence;
  • domain rules and HTTP/CLI details;
  • database schema models and public API schemas unless deliberate;
  • configuration parsing and runtime behavior;
  • hardware control and UI state;
  • integration retries and domain decisions.

When boundaries are already messy, make the change without making them worse.

Domain-Driven Design

Use DDD where domain complexity justifies it.

Prefer DDD concepts when the codebase has:

  • rich business rules;
  • multiple domain concepts with invariants;
  • workflows that need clear language;
  • bounded contexts;
  • domain events or policies;
  • complex state transitions.

Do not use DDD ceremony for simple data plumbing.

DDD Practical Rules

  • Use domain language in names.
  • Keep invariants close to the domain model.
  • Make invalid states hard or impossible to represent.
  • Keep application orchestration separate from domain decisions.
  • Keep infrastructure adapters outside the domain model.
  • Keep bounded contexts explicit.
  • Do not leak persistence concerns into domain behavior.

Clean Architecture Practical Rules

  • Business rules should not know about frameworks.
  • Use cases should express application behavior.
  • Adapters translate between external systems and internal models.
  • Interface boundaries should exist where they reduce coupling, not because diagrams enjoy arrows.
  • Dependency inversion is useful when there are multiple implementations, test seams, or infrastructure volatility.

Package / Module Design

A package should have a reason to exist.

Good package reasons:

  • one bounded context;
  • one application capability;
  • one integration boundary;
  • one infrastructure adapter;
  • one presentation area;
  • one hardware responsibility.

Bad package reasons:

  • "misc";
  • "common" full of unrelated code;
  • "services" as a junk drawer;
  • "models" containing every shape in the universe.

Interface Rules

Introduce an interface/abstraction when it helps with:

  • dependency inversion;
  • testability;
  • multiple implementations;
  • external system isolation;
  • stable domain/application contracts.

Do not introduce interfaces for one implementation unless the boundary is meaningful.

Data Flow

Make data flow explicit.

Avoid:

  • hidden service locators;
  • global mutable registries;
  • implicit configuration reads everywhere;
  • side-effectful imports;
  • action-at-a-distance state changes.

Architecture Review Checklist

Before finishing an architectural change, verify:

  • responsibilities are cohesive;
  • dependency direction is preserved;
  • public contracts are stable or intentionally changed;
  • new abstractions have real callers or clear boundary value;
  • errors cross boundaries in a controlled way;
  • tests cover important behavior;
  • no layer gained unrelated responsibilities.

Extended References

For deeper architecture work, read the relevant documents:

  • ../../engineering/05_design_principles.md
  • ../../engineering/06_responsibility_model.md
  • ../../engineering/09_architectural_smells.md
  • ../../engineering/22_evolution_rules.md

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.