Skills
skills/hassanhabib/the-standard-skills/The Standard Architecture

The Standard Architecture

Installation
SKILL.md

The Standard Architecture

What this skill is

This skill operationalizes the architecture chapters of The Standard. It governs how systems are decomposed, how responsibilities are assigned, and how dependencies flow. It includes the architecture of models, brokers, services, aggregation, exposers, communication protocols, APIs, and user interfaces.

Explicit coverage map

This skill explicitly covers:

  • 0.1.2 Modeling

    • Data Carrier Models
      • Primary Models
      • Secondary Models
      • Relational Models
      • Hybrid Models
    • Operational Models
      • Integration Models (Brokers)
      • Processing Models (Services)
      • Exposure Models (Exposers)
    • Configuration Models

  • 1 Brokers

    • Introduction
    • On The Map
    • Characteristics
      • Implements a Local Interface
      • No Flow Control
      • No Exception Handling
      • Own Their Configurations
      • Natives from Primitives
      • Naming Conventions
      • Language
      • Up & Sideways
      • One Resource, One Broker
    • Organization
    • Broker Types
      • Entity Brokers
      • Support Brokers
    • Implementation
      • Asynchronization Abstraction
    • FAQs / Clarifications
    • Standardized Provider Abstraction Libraries (SPAL)
      • Extensibility
      • Configurability
      • Distributability
      • External Mockability (Cloud-Foreign)
      • Local to Global

  • 2 Services

    • Introduction
    • Services Operations
      • Validations
      • Processing
      • Integration
    • Services Types
      • Validators
      • Orchestrators
      • Aggregators
    • Overall Rules
      • Do or Delegate
      • Two-Three (Florance Pattern)
      • Single Exposure Point
      • Same or Primitives I/O Model
      • Every Service for Itself
      • Flow Forward
      • For APIs
    • Foundation Services (Broker-Neighboring)
      • Introduction
      • On The Map
      • Characteristics
        • Pure-Primitive
        • Single Entity Integration
        • Business Language
      • Responsibilities
        • Abstraction
        • Validation
          • Circuit-Breaking Validations
          • Continuous Validations
          • Upsertable Exceptions
          • Dynamic Rules
          • Rules & Validations Collector
          • Hybrid Continuous Validations
          • Structural Validations
          • Logical Validations
          • External Validations
          • Dependency Validations
        • Mapping
          • Non-Local Models
        • Exception Handling
          • Exceptions Mappings
            • Localise External Exceptions
            • Caragorize Exceptions
          • Logging
    • Processing Services (Higher-Order Business Logic)
      • Introduction
      • On The Map
      • Characteristics
        • Language
        • Functions Language
        • Pass-Through
        • Class-Level Language
        • Dependencies
        • One-Foundation
        • Used-Data-Only Validations
      • Responsibilities
        • Higher-Order Logic
        • Shifters
        • Combinations
        • Signature Mapping
        • Non-Exception Local Models
        • Exception Handling
          • Unwrap and Localise Foundation Exceptions
          • Caragorize Exceptions
          • Logging
    • Orchestration Services (Complex Higher Order Logic)
      • Introduction
      • On The Map
      • Characteristics
        • Language
        • Functions Language
        • Pass-Through
        • Class-Level Language
        • Dependencies
        • Dependency Balance (Florance Pattern)
        • Two-Three
        • Full-Normalization
        • Semi-Normalization
        • No-Normalization
        • Meaningful Breakdown
        • Contracts
          • Physical Contracts
          • Virtual Contracts
        • Cul-De-Sac
      • Responsibilities
        • Advanced Logic
        • Flow Combinations
        • Call Order
          • Natural Order
          • Enforced Order
        • Exception Handling
          • Unwrap and Localise Foundation Exceptions
          • Caragorize Exceptions
          • Logging
      • Variations
        • Coordination
        • Management
        • Uber Management
        • Unit of Work
    • Aggregation Services (The Knot)
      • Introduction
      • On The Map
      • Characteristics
        • No Dependency Limitation
        • No Order Validation
        • Basic Validations
        • Pass-Through
        • Optionality
        • Routine-Level Aggregation
        • Pure Dependency Contracts
      • Responsibilities
        • Abstraction
        • Exceptions Aggregation

  • 3 Exposers

    • Introduction
    • Purpose
    • Pure Mapping
    • Types of Exposure Components
      • Communication Protocols
      • User Interfaces
      • I/O Components
    • Single Point of Contact
    • Summary

  • 3.1 Communication Protocols

    • Principles & Rules
      • Results Communication
      • Error Reports
    • RESTful APIs
      • Language
      • Beyond CRUD Routines
      • Similar Verbs
      • Route Conventions
        • Nouns & Verbs
        • Controller Routes
        • Routine/Method-Specific Routes
        • Plural-Singular-Plural
        • Query Parameters & OData
        • X-WWW-Form-UrlEncoded Parameters
      • Codes & Responses
        • Success Codes (2xx)
        • User Error Codes (4xx)
        • System Error Codes (5xx)
        • All Codes
      • Single Dependency
      • Single Contract
      • Organization
      • Home Controller

  • 3.2 User Interfaces

    • Principles & Rules
      • Progress (Loading)
        • Basic Progress
        • Remaining Progress
        • Detailed Progress
      • Results
        • Simple
        • Partial Details
        • Full Details
      • Error Reports
        • Informational
        • Referential / Implicit Actions
        • Actionable
      • Single Dependency
      • Anatomy
        • Bases
        • Components
        • Containers
      • UI Component Types
        • Web Applications
        • Mobile Applications
        • Other Types
      • Web Applications
      • Anatomy
        • Base Component
          • Implementation
          • Utilization
          • Restrictions
        • Core Component
          • Elements
            • Existence
              • Property Assignment
              • Searching by Id
              • General Search
            • Properties
            • Actions
          • Styles
          • Actions
          • Restrictions
        • Pages
        • Unobtrusiveness
        • Organization

    The following topics are governed by dedicated skills that extend this skill. See ## Related skills for activation guidance:

    • CulDeSac event brokers, event services, publish/subscribe, DI registration, and startup activation → the-standard-events
    • Release versioning, file versioning, API versioning, and deprecation → the-standard-versioning

When to use

Use this skill for system design, architecture review, decomposition, refactoring, API design, UI architecture, dependency-flow design, or when deciding where behavior belongs.

Related skills

This skill defines the structural rules for the entire system. When a specific architectural pattern is chosen, a dedicated skill governs its detailed implementation. Activate the relevant skill as soon as the pattern is identified -- do not wait until implementation is underway.

Pattern Skill Activate when
CulDeSac eventing: event broker, event service, publish/subscribe, DI registration, startup activation the-standard-events An orchestration service needs to publish a domain event or subscribe to one without a synchronous return
Release versioning, file versioning, API versioning, deprecation the-standard-versioning A model, service, or API contract change is introduced, a release is being cut, or existing code must be deprecated

System-wide architecture rules

  1. Every system must be decomposed into purpose, dependency, and exposure.
  2. At low level, that means:
    • Brokers = dependencies
    • Services = purpose
    • Exposers = exposure
  3. Keep the flow forward:
    • Exposer -> Service -> Broker -> External resource
  4. Never reverse the flow.
  5. Never blur layer boundaries.
  6. Never hide architecture behind magical abstractions.

Modeling rules

Data-carrier models

  1. Primary models are pillars of the system.
    • They are physically self-sufficient.
  2. Secondary models depend on primary models.
  3. Relational models connect primary models and should mainly hold references.
  4. Hybrid models are allowed only when a relationship itself must carry state or details.
  5. Keep physical models anemic and flat.
  6. Use virtual contracts only when orchestration truly needs them.

Operational models

  1. Integration models are brokers.
  2. Processing models are services.
  3. Exposure models are exposers.
  4. Configuration models compose the system and manage startup, DI, middleware, or platform-specific configuration.

Broker rules

  1. A broker is a liaison between business logic and the outside world.
  2. Brokers must implement a local interface.
  3. Brokers must contain no business logic.
  4. Brokers must contain no flow control.
    • No if statements for business decisions.
    • No loops for business decisions.
    • No switch cases for business decisions.
  5. Brokers must not handle exceptions.
    • Let native exceptions propagate to broker-neighboring services.
  6. Brokers must own their configurations.
  7. Brokers may construct native models from primitive inputs.
  8. Brokers must speak the language of their technology.
    • Storage -> Insert / Select / Update / Delete
    • Queue -> Enqueue / Dequeue
    • REST -> Get / Post / Put / Delete / custom HTTP method when needed
  9. Brokers cannot call other brokers.
  10. Brokers cannot depend on services.
  11. One resource, one broker.
  12. Use support brokers for generic capabilities such as time and logging.
  13. Use entity brokers / api brokers for resource- or entity-specific integrations.
  14. Prefer partial interfaces and partial classes to organize multi-entity brokers.
  15. Prefer generic helper methods in broker root partials so entity partials do not touch native clients directly.
  16. Use asynchronous abstractions consistently.
  17. Prefer ValueTask in Standard examples and abstractions when that aligns with the implementation profile.
  18. Brokers live under Brokers/ and their namespaces.
  19. Broker configurations live in appsettings.json or equivalent configuration stores
  20. Broker configuration classes live under Brokers/ and their namespaces.

Asynchronization Abstraction (§1.5.1)

Every publicly exposed interface method — on brokers, services, and exposers — must return ValueTask or ValueTask<T>, even if the current implementation does not internally await anything.

This is The Standard §1.5.1: the public contract is uniformly async so callers never need to change if an implementation later becomes truly asynchronous.

Pattern Verdict
public async ValueTask LogWarningAsync(string message) => this.logger.LogWarning(message); Correct — async keyword, direct call
public ValueTask<IQueryable<Student>> SelectAllStudentsAsync() => ValueTask.FromResult(...) Correct — ValueTask.FromResult wraps sync result
public async ValueTask<IQueryable<Student>> SelectAllStudentsAsync() => await this.SelectAllAsync<Student>(); Correct — async/await through generic helper
public IQueryable<Student> SelectAllStudents() => this.Students.AsNoTracking(); WRONG — synchronous, no ValueTask
public ValueTask LogWarningAsync(string message) => new ValueTask(Task.Run(() => ...)); WRONG — Task.Run wraps a synchronous call needlessly

Consequence for services: CreateAndLog* helpers must be async too, because ILoggingBroker.LogErrorAsync returns ValueTask. Catch blocks must use throw await:

// WRONG
private StudentValidationException CreateAndLogValidationException(Xeption exception)
{
    ...
    this.loggingBroker.LogError(studentValidationException);  // no such sync method
    return studentValidationException;
}

// CORRECT
private async ValueTask<StudentValidationException> CreateAndLogValidationException(
    Xeption exception)
{
    var studentValidationException = new StudentValidationException(...);
    await this.loggingBroker.LogErrorAsync(studentValidationException);
    return studentValidationException;
}

// CORRECT call site in TryCatch
catch (NullStudentException nullStudentException)
{
    throw await CreateAndLogValidationException(nullStudentException);
}

Broker clarifications

  1. Brokers are broader than repositories.
  2. Providers are not the same as brokers.
  3. Native exceptions may leak through brokers by design; foundation services localize them.
  4. Partialization is preferred for multi-entity brokers because configuration ownership stays centralized.
  5. Suppress warnings at the project level when truly needed; otherwise fix them.

SPAL rules

  1. Standardized Provider Abstraction Libraries must be extensible.
  2. They must be configurable.
  3. They must be distributable.
  4. They must support external mockability for local / airplane-mode operation.
  5. They should move from local need to global reusable library when possible.
  6. They are subsystems and should themselves follow Brokers / Services / Exposers.

Service-wide rules

  1. Services contain business logic.
  2. Service operations break into validations, processing, and integration.
  3. Service types break into validators, orchestrators, and aggregators.
  4. Do or delegate, but not both.
  5. Enforce Florance Pattern where applicable.
  6. Exposure layers must have a single point of contact with business logic.
  7. Services should accept and return the same contract or primitives/aggregations of that contract.
    • Methods that has primitive inputs must consider using a contract model count exceed three.
  8. Every service validates its own inputs and outputs.
  9. Services cannot call other services at the same level.
  10. Service methods cannot call other service methods at the same level.
    • If shared logic exists, extract it to a private method that both public methods can call.
  11. Public APIs cannot call public APIs at the same level.
  12. Flow forward only.

Foundation service rules

  1. Foundation services are broker-neighboring services.
  2. Their purpose is validation, abstraction, mapping, and primitive business language.
  3. They must remain pure-primitive.
    • No multi-step higher-order business logic.
  4. They must integrate with one entity broker only.
  5. Support brokers like logging and time are allowed.
  6. They must translate technology language to business language.
    • Insert -> Add
    • Select -> Retrieve
    • Update -> Modify
    • Delete -> Remove
  7. They must wrap logic in TryCatch / exception-noise-cancellation.
  8. They must perform validation before dependency calls.
  9. They are the last abstraction layer before core business logic.
  10. Foundation services live under Services/Foundations.
  11. Fondation service models live under Models/Foundations/{Entity Plural}/{Entity}.cs
  12. Fondation service exceptionmodels live under Models/Foundations/{Entity Plural}/Exceptions/

Validation rules

  1. Validation order is mandatory:
    • Structural
    • Logical
    • External
  2. Circuit-breaking validations exit immediately.
  3. Continuous validations accumulate errors, then break the circuit.
  4. Use upsertable exceptions for accumulated validation data.
  5. Use dynamic rules that include both condition and human-readable message.
  6. Use a rule collector to aggregate and then throw.
  7. Use hybrid continuous validation for nested models.
  8. Validate outgoing data when the current routine uses it.
  9. Map native failures into local exception models.

Exception rules for foundation services

  1. Localize native exceptions.
    • The data dictionary from native exception must be assigned to the localised exception's data dictionary.
    • The inner exception of the localised exception must be the native exception when possible.
  2. Categorize exceptions into validation, dependency validation, dependency, and service exceptions.
  3. Preserve inner localized exceptions when moving upstream.
  4. Log at the appropriate severity.
  5. Critical dependency failures are infrastructure/configuration failures.
  6. Do not leak raw native exception semantics into upstream pure business logic.

Processing service rules

  1. Processing services hold higher-order single-entity business logic.
  2. They may combine primitive operations from one foundation service.
  3. They may use utility brokers.
  4. They may not use entity brokers directly.
  5. They may depend on one and only one foundation service.
  6. Their naming must include the entity and the Processing suffix.
  7. They validate only what they use from the input.
  8. They may shift outcomes to primitives such as bool or int.
  9. They may combine multiple primitive routines into one higher-order routine.
  10. They map exceptions from foundation layer to processing-layer exception categories.
  11. They must localise and categorise exceptions from the foundation layer.
  12. Processing services live under Services/Processings.
  13. Processing service virtual models live under Models/Processings/{Entity Plural}/{Entity}.cs
  14. Processing service exceptionmodels live under Models/Processings/{Entity Plural}/Exceptions/

Orchestration service rules

  1. Orchestration services combine multi-entity operations.
  2. They may depend on foundation services or processing services, but not a mixed set of both for entity/business dependencies.
  3. Utility brokers are allowed in addition.
  4. Florance Pattern is mandatory.
  5. Two-Three rule is mandatory for entity/business service dependencies.
  6. If dependencies exceed the rule, normalize.
    • Full normalization
    • Semi-normalization
    • No-normalization only as the last option
  7. Every breakdown must be meaningful.
  8. Orchestration services may be pass-through when contract purity is preserved.
  9. Orchestration services may expose physical contracts or virtual contracts.
  10. They are responsible for mapping and branching.
  11. They are responsible for calling dependencies in the proper order.
  12. Natural order is preferred when dependencies require it.
  13. Enforced order must be tested when the call dependency is not naturally encoded in input/output relationships.
  14. Cul-De-Sac orchestration is valid for event/listener scenarios.
  15. Variants include coordination, management, and uber-management services.
  16. Keep a unit-of-work mindset.
  17. Prefer eventing when it reduces orchestration complexity safely.
  18. They map exceptions from foundation layer or processing layer to orchestration-layer exception categories.
  19. They must localise and categorise exceptions from the foundation layer or processing layer.
  20. Orchestration services live under Services/Orchestrations.
  21. Orchestration service virtual models live under Models/Orchestrations/{Entity Plural}/{Entity}.cs
  22. Orchestration service exceptionmodels live under Models/Orchestrations/{Entity Plural}/Exceptions/
  23. When Cul-De-Sac eventing is chosen (rules 13 and 16), activate the-standard-events skill -- it governs the event broker, event service, validation, exception handling, DI lifetime, and startup activation for the event layer.

Aggregation service rules

  1. Aggregation services are the knot at the border of core business logic.
  2. They provide a single exposure point when many same-variation services share the same contract.
  3. They do not add business logic.
  4. They can have many same-variation dependencies.
  5. They do not validate call order.
  6. They only perform basic structural validations.
  7. They may aggregate by multi-call routine or by pass-through methods.
  8. They are optional.
  9. Their dependencies must share the same contract family.
  10. They must aggregate exceptions the same way orchestration-like services do.
  11. They must localise and categorise exceptions the same way orchestration-like services do.
  12. Aggregation services lives under Services/Aggregations.
  13. Aggregation service exceptionmodels live under Models/Aggregations/{Entity Plural}/Exceptions/

Exposer rules

  1. Exposers are disposable mapping layers.
  2. Their purpose is duplex mapping in and out of the core business logic.
  3. They are pure mapping only.
  4. They may not talk to brokers.
  5. They may not contain business logic.
  6. They may have one and only one service dependency.
  7. That dependency must be a service, not a broker.
  8. They must provide a single point of contact.

Exposure component types

  1. Communication protocols
  2. User interfaces
  3. I/O components / background daemons

Communication protocol rules

  1. Return core-business results in the protocol’s form.
  2. Return error reports faithfully and with standardized codes.
  3. Support result communication and error reporting.
  4. Keep mapping thin and explicit.

RESTful API rules

  1. Controllers speak HTTP verb language.
    • Post / Get / Put / Delete
  2. Custom verbs are allowed when the business operation goes beyond CRUD and the standard verbs do not fit.
  3. Similar verbs are allowed across different routines when names and routes differentiate them.
  4. Routes must never contain verbs.
  5. Routes must use nouns.
  6. Enforce the single-noun principle.
  7. Prefer resource intersections to noun-collisions.
  8. Controller classes are plural.
  9. Controller routes should usually follow api/[controller].
  10. Method-specific routes extend the controller route.
  11. Follow plural-singular-plural route patterns for nested resource intersections.
  12. Use query parameters and OData where appropriate for queryable reads.
  13. x-www-form-urlencoded is allowed for form-style endpoint inputs.
  14. Controller methods should not accept more than three parameters; beyond that, design a model.
  15. Controllers have one service dependency.
  16. Controllers honor a single contract family.
  17. Controllers live under Controllers.
  18. Every system should have a HomeController heartbeat endpoint.
  19. HomeController should not require security and should only indicate aliveness.

REST response code rules

  1. Success responses:
    • 200 OK for successful GET / PUT / DELETE style outcomes
    • 201 Created for successful POSTs
    • 202 Accepted for delegated or eventual-consistency submissions
  2. User error responses:
    • 400 BadRequest for validation and dependency-validation issues when mapped to user-correctable input/domain problems
    • 404 NotFound for not-found scenarios
    • 409 Conflict for already-exists scenarios
    • 423 Locked for locked-resource scenarios
    • 424 FailedDependency for invalid-reference scenarios
  3. System error responses:
    • 500 InternalServerError for dependency or service failures
    • 507 InsufficientStorage for internal storage issues when applicable
  4. Preserve security boundaries.
    • Do not expose internal details from dependency/service failures unless the protocol requires a sanitized representation.

User interface rules

  1. UI exposers must map progress, results, and errors.
  2. Never fake progress.
  3. Choose progress reporting level intentionally:
    • Basic progress
    • Remaining progress
    • Detailed progress
  4. Choose results reporting level intentionally:
    • Simple
    • Partial details
    • Full details
  5. It is a violation to redirect users after submission with no indication of what happened.
  6. Error reports must tell users what happened, why it happened, and the next course of action when possible.
  7. Error report types:
    • Informational
    • Referential / implicit action
    • Actionable
  8. Translate technical error language into user-appropriate language.
  9. UI exposers have one single dependency.
  10. UI anatomy must separate:
    • Bases
    • Components
    • Containers
  11. Base components wrap native or third-party UI elements.
  12. Core components integrate with one and only one view service.
  13. Containers/pages orchestrate UI components and routes only.
  14. Containers may not contain UI logic.
  15. Containers may not use base components directly.
  16. Base components may not wrap more than one non-local component when avoidable.
  17. Base components do not contain business logic, exception handling, validations, or calculations.
  18. Pages are route containers and do not require business logic.
  19. Separate markup, code-behind, and style files.
  20. Organize UI under Views/Bases, Views/Components, and Views/Pages.
  21. Use domain-driven UI organization.

Web-application specifics

  1. Base components behave like brokers.
  2. Core components behave like service/controller hybrids.
  3. Core components are test-driven.
  4. Core components are built from:
    • Elements
    • Styles
    • Actions
  5. Element testing must cover:
    • Existence
    • Properties
    • Actions
  6. Existence testing may use:
    • Property assignment
    • Searching by id
    • General search
  7. Styles belong primarily to core components, not bases.
  8. Pages compose components and represent routes.
  9. Pages typically do not require unit tests.
  10. Keep UI unobtrusive; do not place CSS, C#, and markup in the same file.
  11. Core components do not call other core components at the same level.
  12. One view service corresponds to one core component and one view model.

Versioning and breaking changes

This skill governs what the correct structure is at each layer. When a structural change breaks an existing contract -- a model property added or removed, a service signature changed, an API route or response shape altered -- it is no longer purely an architecture decision. Activate the-standard-versioning skill at that point.

The versioning skill governs:

  • When and how to increment the release version
  • How to introduce new model versions under Vn subfolders without overwriting earlier versions
  • How to introduce new service versions alongside earlier ones
  • How to version API routes so earlier consumers are not broken
  • How to signal deprecation on models, services, and routes

The architecture skill and the versioning skill are complementary:

  • Architecture answers: is the shape correct?
  • Versioning answers: how do we change it safely?

Architecture review checklist

When reviewing or generating architecture, verify all of the following:

  1. Purpose is explicit.
  2. Models are scoped to the purpose.
  3. Dependencies and exposure are separate from purpose.
  4. Broker responsibilities are thin and disposable.
  5. Services own business logic.
  6. Validation order is correct.
  7. Service-layer flow is forward only.
  8. Same-level services are not calling each other.
  9. Florance Pattern is honored.
  10. Exposure layers have single point of contact.
  11. APIs honor route and status-code rules.
  12. UI honors single dependency, unobtrusiveness, and anatomy rules.
  13. Architecture remains readable, autonomous, and rewritable.
  14. If Cul-De-Sac eventing is used, the-standard-events skill has been activated and its rules are satisfied.
  15. If a model, service, or API contract has changed or a release is being cut, the-standard-versioning skill has been activated and its rules are satisfied.

.NET implementation profile included from the supplied implementation specification

The following addendum preserves the supplied implementation profile so the skill can enforce both the abstract Standard and the concrete .NET implementation style you attached.

1. Overview

The Standard is an opinionated software engineering standard authored by Hassan Habib. It prescribes a tri-nature architecture consisting of:

Layer Responsibility
Brokers Thin abstraction over any external dependency (DB, API, logs)
Services All business logic — validation, orchestration, coordination
Exposers Entry points that expose services (Controllers, Endpoints)

This project currently implements Brokers and Foundation Services for the LegacyUser entity (storage-based) and the Person entity (API-based), as well as an API Broker for sending Person entities to an external API.

2. Project Structure

RedRhino.Core.Synchronizer/
├── Brokers/
│   ├── Apis/
│   │   ├── IModernApiBroker.cs             (partial interface — base)
│   │   ├── IModernApiBroker.Persons.cs     (partial interface — entity)
│   │   ├── ModernApiBroker.cs              (partial class — base)
│   │   └── ModernApiBroker.Persons.cs      (partial class — entity)
│   ├── Loggings/
│   │   ├── ILoggingBroker.cs
│   │   └── LoggingBroker.cs
│   └── Storages/
│       ├── IStorageBroker.cs              (partial interface — base)
│       ├── IStorageBroker.LegacyUsers.cs  (partial interface — entity)
│       ├── StorageBroker.cs               (partial class — base)
│       └── StorageBroker.LegacyUsers.cs   (partial class — entity)
├── Models/
│      ├── Foundations/
│            ├── Persons/
│            │   ├── Person.cs
│            │   ├── PersonType.cs
│            │   ├── PersonRecordState.cs
│            │   └── Exceptions/
│            │       ├── NullPersonException.cs
│            │       ├── InvalidPersonException.cs
│            │       ├── AlreadyExistsPersonException.cs
│            │       ├── FailedPersonDependencyException.cs
│            │       ├── FailedPersonServiceException.cs
│            │       ├── PersonValidationException.cs
│            │       ├── PersonDependencyException.cs
│            │       ├── PersonDependencyValidationException.cs
│            │       └── PersonServiceException.cs
│            └── LegacyUsers/
│                ├── LegacyUser.cs
│                └── Exceptions/
│                    ├── NullLegacyUserException.cs
│                    ├── InvalidLegacyUserException.cs
│                    ├── AlreadyExistsLegacyUserException.cs
│                    ├── FailedStorageLegacyUserDependencyException.cs
│                    ├── FailedLegacyUserServiceException.cs
│                    ├── LegacyUserValidationException.cs
│                    ├── LegacyUserDependencyException.cs
│                    ├── LegacyUserDependencyValidationException.cs
│                    └── LegacyUserServiceException.cs
├── Services/
│   └── Foundations/
│       ├── LegacyUsers/
│       │   ├── ILegacyUserService.cs
│       │   ├── LegacyUserService.cs               (partial — logic)
│       │   ├── LegacyUserService.Validations.cs   (partial — validations)
│       │   └── LegacyUserService.Exceptions.cs    (partial — exception handling)
│       └── Persons/
│           ├── IPersonService.cs
│           ├── PersonService.cs                    (partial — logic)
│           ├── PersonService.Validations.cs        (partial — validations)
│           └── PersonService.Exceptions.cs         (partial — exception handling)
└── Program.cs

RedRhino.Core.Synchronizer.Tests.Units/
└── Services/
    └── Foundations/
        ├── LegacyUsers/
        │   ├── LegacyUserServiceTests.cs              (partial — setup & helpers)
        │   ├── LegacyUserServiceTests.Logic.{Method}.cs        (partial — happy-path tests)
        │   ├── LegacyUserServiceTests.Validations.{Method}.cs  (partial — validation tests)
        │   └── LegacyUserServiceTests.Exceptions.{Method}.cs   (partial — exception tests)
        └── Persons/
            ├── PersonServiceTests.cs                   (partial — setup & helpers)
            ├── PersonServiceTests.Logic.{Method}.cs             (partial — happy-path tests)
            ├── PersonServiceTests.Validations.{Method}.cs       (partial — validation tests)
            └── PersonServiceTests.Exceptions.{Method}.cs        (partial — exception tests)

3. Brokers

Brokers are thin wrappers around external resources. They contain zero business logic.

Rule — Generic Helpers: Every broker base partial must expose private generic helper methods (e.g., InsertAsync<T>, PostAsync<T>) that encapsulate the underlying client calls. Entity partial files never reference the private client member (e.g., this.apiClient, this.Entry(...)) directly — they delegate to the generic helpers instead. This keeps entity partials decoupled from the concrete client and allows swapping the underlying implementation in a single place.

3.1 Storage Broker

Aspect Implementation
Base class EFxceptionsContext (from the EFxceptions library — wraps DbContext with meaningful EF exceptions)
Interface partial interface IStorageBroker — split per entity
Class partial class StorageBroker — split per entity
Generic CRUD helpers Private helpers in base partial: InsertAsync<T>, SelectAllAsync<T>, SelectAsync<T>, UpdateAsync<T>, DeleteAsync<T>
Configuration Reads DefaultConnection from IConfiguration; calls this.Database.Migrate() at construction

Base partial — StorageBroker.cs

This file owns: IConfiguration, OnConfiguring, the constructor (with Database.Migrate()), and private generic CRUD helpers. It owns nothing else.

arch-014 — No DbSet<> in the base partial. DbSet<Student> Students lives in StorageBroker.Students.cs, not here. The base partial must never declare entity-specific members.

public partial class StorageBroker : EFxceptionsContext, IStorageBroker
{
    // No DbSet<> properties here. Each entity partial declares its own.

    private readonly IConfiguration configuration;

    public StorageBroker(IConfiguration configuration)
    {
        this.configuration = configuration;

        // arch-011: Must be called — applies pending migrations at startup.
        // Omitting this means the schema is never applied.
        this.Database.Migrate();
    }

    protected override void OnConfiguring(DbContextOptionsBuilder optionsBuilder)
    {
        string connectionString =
            this.configuration.GetConnectionString("DefaultConnection");

        optionsBuilder.UseSqlServer(connectionString);
    }

    // arch-012: All EF operations live here. Entity partials call these helpers only.
    private async ValueTask<T> InsertAsync<T>(T entity) where T : class
    {
        this.Entry(entity).State = EntityState.Added;
        await this.SaveChangesAsync();
        return entity;
    }

    // SelectAllAsync<T> wraps IQueryable in a ValueTask so all entity methods
    // follow the uniform async/await pattern. AsNoTracking() is applied here.
    private ValueTask<IQueryable<T>> SelectAllAsync<T>() where T : class =>
        ValueTask.FromResult<IQueryable<T>>(this.Set<T>().AsNoTracking());

    private async ValueTask<T> SelectAsync<T>(Guid entityId) where T : class =>
        await this.FindAsync<T>(entityId);

    private async ValueTask<T> UpdateAsync<T>(T entity) where T : class
    {
        this.Entry(entity).State = EntityState.Modified;
        await this.SaveChangesAsync();
        return entity;
    }

    private async ValueTask<T> DeleteAsync<T>(T entity) where T : class
    {
        this.Entry(entity).State = EntityState.Deleted;
        await this.SaveChangesAsync();
        return entity;
    }
}

Entity partial — StorageBroker.Students.cs

arch-014: DbSet<Student> is declared here — in the entity partial — not in StorageBroker.cs.

public partial class StorageBroker
{
    // DbSet<Student> belongs in this entity partial file, not in StorageBroker.cs.
    public DbSet<Student> Students { get; set; }

    public async ValueTask<Student> InsertStudentAsync(Student student) =>
        await this.InsertAsync(student);

    // arch-009: SelectAll* must be async ValueTask<IQueryable<T>>.
    // WRONG:   public IQueryable<Student> SelectAllStudents() => this.Students.AsNoTracking();
    // WRONG:   public IQueryable<Student> SelectAllStudents() => this.Set<Student>();
    // CORRECT: delegate to SelectAllAsync<T>() — never touch DbSet or EF members directly.
    public async ValueTask<IQueryable<Student>> SelectAllStudentsAsync() =>
        await this.SelectAllAsync<Student>();

    public async ValueTask<Student> SelectStudentByIdAsync(Guid studentId) =>
        await this.SelectAsync<Student>(studentId);

    public async ValueTask<Student> UpdateStudentAsync(Student student) =>
        await this.UpdateAsync(student);

    public async ValueTask<Student> DeleteStudentAsync(Student student) =>
        await this.DeleteAsync(student);
}

Rule: Each entity gets its own partial file for both the interface and the class.

Rule — scope before scaffolding (arch-013): A branch named BROKERS-student-insert implements only InsertStudentAsync. If the prompt is ambiguous about which operations are needed, the agent must ask before creating the branch or writing any code. A single broker branch = a single operation.

Rule — branch action language (prac-016): Broker branch actions use infrastructure verbs — insert, select-all, select-by-id, update, delete. Never use business verbs (add, retrieve, modify, remove) in a BROKERS branch name.

3.2 API Broker

Aspect Implementation
HTTP client IRESTFulApiFactoryClient (from the RESTFulSense library — wraps HttpClient)
Interface partial interface IModernApiBroker — split per entity
Class partial class ModernApiBroker — split per entity
Generic HTTP helpers Private PostAsync<T>() on the base partial for reuse across entities
Configuration Reads ApiConfigurations:Url from IConfiguration to set HttpClient.BaseAddress

Base partial — ModernApiBroker.cs

public partial class ModernApiBroker : IModernApiBroker
{
    private readonly IRESTFulApiFactoryClient apiClient;

    public ModernApiBroker(IConfiguration configuration)
    {
        var httpClient = new HttpClient()
        {
            BaseAddress =
                new Uri(configuration.GetValue<string>("ApiConfigurations:Url"))
        };

        this.apiClient = new RESTFulApiFactoryClient(httpClient);
    }

    private async ValueTask<T> PostAsync<T>(string relativeUrl, T content) =>
        await this.apiClient.PostContentAsync<T>(relativeUrl, content);
}

Entity partial — ModernApiBroker.Persons.cs

public partial class ModernApiBroker
{
    private const string PersonsRelativeUrl = "api/persons";

    public async ValueTask<Person> PostPersonAsync(Person person) =>
        await PostAsync(PersonsRelativeUrl, person);
}

Rule: Entity partials delegate to the generic helpers (PostAsync<T>) — they never call this.apiClient directly.

3.3 Logging Broker

A dedicated abstraction over ILogger<T>. The broker exposes only async ValueTask methods that correspond to standard log levels:

Method Log Level
LogInformationAsync Information
LogTraceAsync Trace
LogDebugAsync Debug
LogWarningAsync Warning
LogErrorAsync Error
LogCriticalAsync Critical

LoggingBroker.cs

public class LoggingBroker : ILoggingBroker
{
    private readonly ILogger<LoggingBroker> logger;

    public LoggingBroker(ILogger<LoggingBroker> logger) =>
        this.logger = logger;

    public async ValueTask LogInformationAsync(string message) =>
        this.logger.LogInformation(message);

    public async ValueTask LogTraceAsync(string message) =>
        this.logger.LogTrace(message);

    public async ValueTask LogDebugAsync(string message) =>
        this.logger.LogDebug(message);

    public async ValueTask LogWarningAsync(string message) =>
        this.logger.LogWarning(message);

    public async ValueTask LogErrorAsync(Exception exception) =>
        this.logger.LogError(exception, exception.Message);

    public async ValueTask LogCriticalAsync(Exception exception) =>
        this.logger.LogCritical(exception, exception.Message);
}

ILoggingBroker.cs

public interface ILoggingBroker
{
    ValueTask LogInformationAsync(string message);
    ValueTask LogTraceAsync(string message);
    ValueTask LogDebugAsync(string message);
    ValueTask LogWarningAsync(string message);
    ValueTask LogErrorAsync(Exception exception);
    ValueTask LogCriticalAsync(Exception exception);
}

Rule — async expression body, no Task.Run: Each method uses the async keyword and delegates directly to ILogger<T>. Never wrap the call in Task.Run() or new ValueTask(Task.Run(...)). ILogger<T> is synchronous; wrapping it in Task.Run() introduces unnecessary thread-pool overhead and produces an inefficient heap-allocated ValueTask.

Wrong:

public ValueTask LogWarningAsync(string message) =>
    new ValueTask(Task.Run(() => this.logger.LogWarning(message)));

Correct:

public async ValueTask LogWarningAsync(string message) =>
    this.logger.LogWarning(message);

Rule — DI Registration: The logging broker must be explicitly registered in Program.cs. AddLogging() (or the host builder's default) must also be present so that ILogger<LoggingBroker> resolves correctly.

builder.Services.AddLogging();
builder.Services.AddTransient<ILoggingBroker, LoggingBroker>();

Omitting either line causes a runtime DI resolution failure.

4. Models

4.1 Entity Model

Entity classes are plain POCOs residing under Models/{Service Type}/{Entity}/. They contain no behavior — only properties. Domain comments on properties capture validation intent (e.g., nullable rules, email format, phone format).

The LegacyUser class resides under Models/Foundations/LegacyUsers/. The Person class resides under Models/Foundations/Persons/.

4.2 Exception Models

Exception models live in Models/{Service Type}/{Entity}/Exceptions/ and form a two-tier exception hierarchy per The Standard.

The LegacyUser class resides under Models/Foundations/LegacyUsers/Exceptions/. The Person class resides under Models/Foundations/Persons/Exceptions/.

The inner/outer exception hierarchy varies by the type of broker the Foundation Service consumes. Storage-based services (e.g., LegacyUserService) handle SQL/EF exceptions, while API-based services (e.g., PersonService) handle RESTFulSense HTTP exceptions.

4.2.1 Storage-Based Exceptions (LegacyUser)

Inner (Local) Exceptions
Exception Purpose
NullLegacyUserException Entity is null
InvalidLegacyUserException One or more property-level validation fails
AlreadyExistsLegacyUserException Duplicate key detected
FailedStorageLegacyUserDependencyException SQL / storage-level failure
FailedLegacyUserServiceException Unexpected runtime failure
Outer (Categorical) Exceptions
Exception Category Wrapped Inner(s)
LegacyUserValidationException Validation NullLegacyUserException, InvalidLegacyUserException
LegacyUserDependencyException Dependency FailedStorageLegacyUserDependencyException
LegacyUserDependencyValidationException DependencyValidation AlreadyExistsLegacyUserException, InvalidLegacyUserException (from DbUpdateException)
LegacyUserServiceException Service FailedLegacyUserServiceException

4.2.2 API-Based Exceptions (Person)

Inner (Local) Exceptions
Exception Purpose
NullPersonException Entity is null
InvalidPersonException One or more property-level validation fails, or BadRequest from API
AlreadyExistsPersonException Conflict (409) from API
FailedPersonDependencyException HTTP-level failure (any HTTP error or HttpRequestException)
FailedPersonServiceException Unexpected runtime failure
Outer (Categorical) Exceptions
Exception Category Wrapped Inner(s)
PersonValidationException Validation NullPersonException, InvalidPersonException
PersonDependencyValidationException DependencyValidation InvalidPersonException (from BadRequest), AlreadyExistsPersonException (from Conflict)
PersonDependencyException Dependency (Critical) FailedPersonDependencyException (from Unauthorized, Forbidden, NotFound, UrlNotFound, HttpRequestException)
PersonDependencyException Dependency (Non-Critical) FailedPersonDependencyException (from InternalServerError, ServiceUnavailable)
PersonServiceException Service FailedPersonServiceException

All exceptions derive from Xeption (from the Xeption NuGet package), which provides UpsertDataList and ThrowIfContainsErrors for aggregated validation data.

5. Services — Foundations

A Foundation Service sits directly on top of Brokers and is the only consumer of them.

5.1 Partial Class Layout

The service is split into three partial files:

Partial file Concern
{Entity}Service.cs Constructor, DI fields, public business logic
{Entity}Service.Validations.cs All Validate* and IsInvalid* methods
{Entity}Service.Exceptions.cs TryCatch delegate pattern, CreateAndLog* helpers

5.2 Dependency Injection

A Foundation Service depends only on Brokers — never on other services.

Storage-based service (LegacyUserService):

public partial class LegacyUserService : ILegacyUserService
{
    private readonly IStorageBroker storageBroker;
    private readonly ILoggingBroker loggingBroker;

    public LegacyUserService(
        IStorageBroker storageBroker,
        ILoggingBroker loggingBroker) { ... }
}

API-based service (PersonService):

public partial class PersonService : IPersonService
{
    private readonly IModernApiBroker modernApiBroker;
    private readonly ILoggingBroker loggingBroker;

    public PersonService(
        IModernApiBroker modernApiBroker,
        ILoggingBroker loggingBroker) { ... }
}

5.3 Business Logic — TryCatch Pattern

Every public method delegates to a TryCatch wrapper that catches, wraps, logs, and re-throws categorised exceptions:

Storage-based:

public ValueTask<LegacyUser> AddLegacyUserAsync(LegacyUser legacyUser) =>
TryCatch(async () =>
{
    ValidateLegacyUser(legacyUser);

    return await this.storageBroker.InsertLegacyUserAsync(legacyUser);
});

API-based:

public ValueTask<Person> AddPersonAsync(Person person) =>
TryCatch(async () =>
{
    ValidatePerson(person);

    return await this.modernApiBroker.PostPersonAsync(person);
});

5.4 Validations

Validations use a dynamic rule + aggregate pattern:

Validate(
    (IsInvalid(legacyUser.Id),        nameof(LegacyUser.Id)),
    (IsInvalid(legacyUser.UserPK),    nameof(LegacyUser.UserPK)),
    (IsInvalid(legacyUser.UserName),  nameof(LegacyUser.UserName)),
    ...);

Each IsInvalid overload returns an anonymous object with Condition (bool) and Message (string). The Validate method aggregates all failures into a single InvalidLegacyUserException via UpsertDataList and calls ThrowIfContainsErrors().

Custom validators exist for domain-specific rules:

Validator Rule
IsInvalid(Guid) Must not be Guid.Empty
IsInvalid(int) Must not be default (0)
IsInvalid(string) Must not be null/empty/whitespace
IsInvalidLob(int) Must be greater than 0
IsInvalidEmail Regex-based email format check

6. Exception Handling

The TryCatch method in each {Entity}Service.Exceptions.cs implements The Standard's exception mapping pattern. The specific catch blocks differ based on the broker type the service consumes.

6.1 Storage-Based Exception Mapping (LegacyUser)

Native / External Exception  →  Inner (Local) Exception       →  Outer (Categorical) Exception
──────────────────────────────────────────────────────────────────────────────────────────────────
(null input)                 →  NullLegacyUserException        →  LegacyUserValidationException
(validation rules fail)      →  InvalidLegacyUserException     →  LegacyUserValidationException
SqlException                 →  FailedStorage...Exception      →  LegacyUserDependencyException       (Critical)
DuplicateKeyException        →  AlreadyExists...Exception      →  LegacyUserDependencyValidationException
DbUpdateException            →  InvalidLegacy...Exception      →  LegacyUserDependencyValidationException
Exception (catch-all)        →  FailedService...Exception      →  LegacyUserServiceException

6.2 API-Based Exception Mapping (Person)

For services that call external APIs through RESTFulSense, the exception mapping covers all HTTP response exceptions. The ordering of catch blocks in TryCatch matters — more specific exceptions must appear before their base classes.

Native / External Exception               →  Inner (Local) Exception           →  Outer (Categorical) Exception         Log Level
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
(null input)                               →  NullPersonException               →  PersonValidationException             LogError
(validation rules fail)                    →  InvalidPersonException             →  PersonValidationException             LogError
HttpResponseBadRequestException            →  InvalidPersonException             →  PersonDependencyValidationException   LogError
HttpResponseConflictException              →  AlreadyExistsPersonException       →  PersonDependencyValidationException   LogError
HttpResponseUnauthorizedException          →  FailedPersonDependencyException    →  PersonDependencyException             LogCritical
HttpResponseForbiddenException             →  FailedPersonDependencyException    →  PersonDependencyException             LogCritical
HttpResponseNotFoundException              →  FailedPersonDependencyException    →  PersonDependencyException             LogCritical
HttpResponseUrlNotFoundException           →  FailedPersonDependencyException    →  PersonDependencyException             LogCritical
HttpResponseInternalServerErrorException   →  FailedPersonDependencyException    →  PersonDependencyException             LogError
HttpResponseServiceUnavailableException    →  FailedPersonDependencyException    →  PersonDependencyException             LogError
HttpRequestException                       →  FailedPersonDependencyException    →  PersonDependencyException             LogCritical
Exception (catch-all)                      →  FailedPersonServiceException       →  PersonServiceException                LogError

Rule — Critical vs Non-Critical Dependency: Exceptions that indicate the API endpoint is unreachable or inaccessible (Unauthorized, Forbidden, NotFound, UrlNotFound, HttpRequestException) are logged at LogCriticalAsync because they signal configuration or infrastructure failures. Server-side errors (InternalServerError, ServiceUnavailable) are logged at LogErrorAsync because they may be transient.

Each CreateAndLog* helper:

  1. Wraps the inner exception in the categorical outer exception.
  2. Logs via the ILoggingBroker at the appropriate level (LogErrorAsync or LogCriticalAsync).
  3. Returns the outer exception so TryCatch can throw it.

7. Dependency Registration (Exposers)

All wiring is done in Program.cs following The Standard's explicit registration style:

builder.Services.AddDbContext<StorageBroker>();
builder.Services.AddTransient<IStorageBroker, StorageBroker>();
builder.Services.AddTransient<IModernApiBroker, ModernApiBroker>();
builder.Services.AddTransient<ILoggingBroker, LoggingBroker>();
builder.Services.AddTransient<ILegacyUserService, LegacyUserService>();
builder.Services.AddTransient<IPersonService, PersonService>();

Brokers and Foundation Services are registered as Transient.

10. Naming Conventions

Element Pattern Example
Broker interface I{Resource}Broker IStorageBroker, IModernApiBroker, ILoggingBroker
Broker class {Resource}Broker StorageBroker, ModernApiBroker, LoggingBroker
Broker method {Action}{Entity}Async InsertLegacyUserAsync, PostPersonAsync
Service interface I{Entity}Service ILegacyUserService
Service class {Entity}Service LegacyUserService
Service method Add{Entity}Async AddLegacyUserAsync
Inner exception {Adjective}{Entity}Exception NullLegacyUserException
Outer exception {Entity}{Category}Exception LegacyUserValidationException
Test class {Entity}ServiceTests LegacyUserServiceTests
Test method Should{Action}Async / ShouldThrow{Exception}On{Action}… ShouldAddLegacyUserAsync
Weekly Installs
–
Repository
hassanhabib/the…d-skills
GitHub Stars
16
First Seen
–
Security Audits
Gen Agent Trust HubPass
SocketPass
SnykPass