Code Standards Dispatcher

When a code task is requested, identify the technology and load the appropriate sub-skill from this directory before generating or reviewing any code.

Routing Table

Instructions

1. Identify the primary technology from the user's request or the existing codebase imports.
2. Load and apply only the matching sub-skill above.
3. If multiple technologies are in scope (e.g. a FastAPI backend + Angular frontend), apply each relevant sub-skill to its respective layer.
4. If the technology is not listed above, apply general clean code and SOLID principles.
5. Any Python task always loads python/SKILL.md first (core language standards). If the task also involves FastAPI, Django REST Framework, or Airflow, additionally load the matching plain reference file — python/frameworks/fastapi.md, python/frameworks/django-rest-framework.md, or python/frameworks/airflow.md — which assumes the base Python standards and only covers what's framework-specific, so load both together. If the task is architectural — domain modeling, aggregates, ports/adapters, a Repository/Unit-of-Work, a composition root, or picking/rejecting a design pattern or DI approach in Python — additionally load python/references/ddd-clean-architecture.md alongside whichever framework reference applies. A Python + Kafka/event-driven task, or a task integrating DDD/TDD/BDD into one workflow, additionally loads python/references/ddd-tdd-bdd-workflow.md.
6. Any TypeScript/JavaScript task always loads typescript-code-standards/SKILL.md first (core language standards — type system idiom, async, modules, tooling). If the task involves Angular, additionally load typescript-code-standards/frameworks/angular.md, which assumes the base TypeScript standards and only covers what's Angular-specific. An Angular task that touches SCSS/CSS, visual layout, or design tokens must additionally load typescript-code-standards/ui-style/SKILL.md (visual design system) — it covers a different concern and is meant to be applied together with the framework reference, not as an alternative. Angular architecture/DDD questions load typescript-code-standards/references/angular-clean-architecture.md alongside the framework reference (see rule 7 below).

7. Language-specific architecture guidance wins over the generic software-architecture sub-skill. When a DDD / SOLID / Clean Architecture / design-pattern question is scoped to a language or platform that ships its own architecture guidance, use that guidance and do not also load the generic software-architecture sub-skill — they diverge on concrete choices (e.g. where repository interfaces live, whether to have one class per use case, function-first vs class-first idiom) and loading both yields contradictory advice. Specifically:

   • Python → python/references/ddd-clean-architecture.md
   • Go → go/SKILL.md (the consumer declares the port, etc.)
   • Rust → rust/SKILL.md
   • R → r/SKILL.md (deliberately adapts/reverses OOP principles for R's functional grain)
   • Swift → swift/SKILL.md (favors value types, enums-with-associated-values, and protocol-witness structs over deep protocol hierarchies — reject porting Java/C#-style layered indirection wholesale)
   • TypeScript + Angular → typescript-code-standards/references/angular-clean-architecture.md (plain TypeScript/Node work with no framework has no dedicated architecture guidance of its own yet — fall through to the generic software-architecture sub-skill for that case)
   • Oracle PL/SQL → database-data-modeling/oracle-plsql/references/clean-architecture.md (bounded contexts as schema boundaries, PL/SQL packages as repository contracts, etc. — do not apply the generic skill's Repository/Aggregate guidance verbatim to Oracle)

   Load the generic software-architecture sub-skill only for language-agnostic architecture questions, or for a language/platform that has no dedicated architecture guidance of its own (PostgreSQL, RabbitMQ, JSON Schema, plain TypeScript/Node with no framework).

8. Oracle tasks do not default to a legacy version. Load database-data-modeling/oracle-plsql/SKILL.md for the version-agnostic PL/SQL core, which covers most naming/exception-handling/bulk-processing/cursor standards regardless of version. Additionally load frameworks/oracle-11g.md only when the request or codebase explicitly targets a legacy/pre-12c deployment — never as a silent default for unversioned work. Additionally load frameworks/oracle-26ai.md when the request or codebase signals Oracle AI Database features: 26ai, VECTOR columns, vector search, DBMS_VECTOR, RAG, or OCI GenAI.
9. For system-level architecture questions (how many services, when to split a monolith, bounded contexts), use the top-level software-architecture-progression skill instead — it governs system topology, while this dispatcher's software-architecture sub-skill governs code structure within a module. They are complementary; apply software-architecture-progression first, then this dispatcher's software-architecture sub-skill.
10. Each language sub-skill carries its own docstring / doc-comment standard (godoc, rustdoc, TSDoc, DocC, roxygen2, PL/SQL, PEP 257). The cross-cutting discipline for documenting production code — when to document, keeping docs in sync with the code (drift), driving the right generator — is the top-level documenting-code skill. Apply the language sub-skill for how a docstring looks; documenting-code for whether/what/when and keeping it accurate.
11. This dispatcher sits in the middle of the delivery lifecycle: it is handed off work from documentation (once a TRD/SAD/ADR settles what to build, this dispatcher governs how the code is written) and hands off in turn to testing for test specification once the code standards have been applied, continuing on to deployment — github-circleci-standards for CircleCI pipelines, zeno-stack/zeno-deployment-patterns for GitHub Actions and for platform-neutral strategy/rollback/readiness.