Conciseness

The body is dense and telegraphic — no concept explanations, just directives — but it duplicates guidance: references/runtime-quality-checklist.md is mandated twice (once as a link in the checklist section, again under 'Project invariants'), both times with 'OBLIGATORIO antes commit y spawn auditor', and the 'Referencias' and 'Project invariants' sections overlap in purpose. This fits anchor 4 ('efficient; minor instances of over-explanation that could be trimmed') rather than 5's 'every token earns its place'.

Actionability

Guidance is fully executable throughout: exact commands ('make extraction-contract', 'cd backend && .venv/bin/pytest tests/architecture/test_extraction_contract.py -x -q', 'make arch-test'), exact paths (docs/domains/INDEX.md, backend/src/modules/{nombre}/), and complete copy-paste code for the SQLAlchemy 2.0 query, Pydantic v2 DTO, and idempotent Alembic migration patterns, each with correct vs. prohibited contrasts. This matches anchor 5's 'fully executable; copy-paste ready code or commands; specific examples cover the common cases'.

Workflow Clarity

The SOP sequences work explicitly: locate module → explore code → task-type procedure, with three distinct procedures (new features inside-out, modifications code-first, bugs outside-in). Validation is built in as explicit RED-first feedback loops ('escribir test que falla (RED) antes de implementar', 'NO corregir hasta tener test rojo'), a mandatory pre-commit runtime checklist, and enforced architecture tests ('El test arquitectural falla si saltas estos pasos'). This matches anchor 5 — clear sequence, explicit validation, feedback loops, and checklists — including for the database/migration operations the rubric flags as needing them.

Progressive Disclosure

The body is a proper overview that pushes detail into references/ ('Referencias (leer solo cuando aplique)' and 'Project invariants (read on demand)'), and all 10 referenced files exist exactly one level deep in the bundle. Two minor gaps keep it from anchor 5: the runtime-quality-checklist link uses a malformed path ('backend-expert/references/runtime-quality-checklist.md' with a stray skill-name prefix instead of a plain relative path), and reference listings are split across two overlapping sections rather than one navigable index.

Specificity

The description lists five concrete, distinct capabilities covering the full backend stack — 'Implements FastAPI endpoints, creates SQLAlchemy 2.0 async models, generates idempotent Alembic migrations, structures bounded contexts following DDD (domain→infrastructure→application→api), and produces typed Pydantic v2 DTOs'. This matches the anchor 'multiple specific concrete actions; comprehensive coverage' — no meaningful gaps remain for this domain.

Completeness

Both questions are answered explicitly: the 'what' is the concrete capability list, and the 'when' is an explicit 'Use when:' clause followed by concrete trigger phrases and a stack catch-all. This is exactly the anchor-5 pattern exemplified by 'Extract text and tables from PDF files... Use when working with PDF files or when the user mentions PDFs'.

Trigger Term Quality

Trigger phrases are natural user utterances with broad synonym coverage in two languages: 'create an endpoint', 'modify the backend', 'create a new entity', 'update a service', 'database logic', 'fix a backend bug', 'add a migration', 'create a repository', 'agrega un campo', 'nueva ruta API', 'corrige el servicio', plus the stack names themselves. This matches the top anchor's 'comprehensive coverage of natural terms including synonyms'.

Distinctiveness Conflict Risk

The niche is clear — this project's FastAPI/SQLAlchemy/Alembic backend inside a named Docker container — and triggers are distinct from frontend/data/admin skills. However the trailing catch-all 'or any Python/FastAPI/SQLAlchemy/Alembic task' is broad enough to grab general Python work unrelated to this backend, so overlap risk is minor rather than minimal — fitting anchor 4, not 5.