santimarquez.es

AI-augmented development

This site is built and maintained with AI-augmented development in mind. The repo is structured so that an AI assistant (e.g. Cursor) can onboard quickly, follow consistent patterns, and run defined workflows for documentation, code review, and SEO. That structure is a deliberate showcase of how to leverage LLMs for faster, consistent output without sacrificing quality—directly applicable to product teams and technical leadership.

AI infrastructure: how the repo is built for AI

Context (ai/context/): Living project knowledge: PROJECT_STRUCTURE (where files go), RECENT_CHANGES (latest decisions), COMMON_PATTERNS (Astro, Tailwind, Motion One), TECH_STACK, CODING_STANDARDS. Single source of truth so any AI or developer can align with conventions.

Skills (ai/skills/): Reusable patterns for common tasks—Astro components, Tailwind sections, Motion One usage, i18n, SVG diagrams. Referenced in chat so generations stay on-brand and consistent.

Agents (ai/agents/): Multi-step workflows (e.g. documentation, code review, SEO checks) that Cursor can run end-to-end.

Prompts (ai/prompts/): Templates for code review, copy transformation (CV to impact statements), documentation, and diagram generation.

Cursor integration: .cursorrules and CURSOR.md (and CURSOR_QUICK_START.md) are loaded by Cursor; they define stack, design system, i18n, file locations, and example prompts. The ai/ folder is the extended context for structure and patterns.

flowchart LR
  Cursor[Cursor] --> Rules[.cursorrules / CURSOR.md]
  Rules --> Context[ai/context]
  Context --> Skills[ai/skills]
  Context --> Agents[ai/agents]
  Context --> Prompts[ai/prompts]
  Skills --> Out[Consistent code and docs]
  Agents --> Out
  Prompts --> Out

Capabilities this demonstrates

Structuring project knowledge for LLMs (context, conventions, patterns); prompt engineering (skills and prompt templates that yield predictable, on-brand output); workflow design (agents for documentation and quality checks); reproducible quality (a11y, SEO, design system) through AI-assisted flows. These are the same capabilities needed to introduce or scale AI-augmented development in a team or product—onboarding, consistency, and guardrails.

What the site delivers

The output is a production-ready static site (Astro, TypeScript, Tailwind, Motion One, i18n EN/ES, Docker, Nginx, Traefik-ready). SEO, Person schema, sitemap, WCAG-oriented a11y. The stack is chosen for performance and simplicity; the AI layer is what keeps development and maintenance efficient and consistent.

Tech stack

Outcome

This repo doubles as a reference for structuring a project for AI-augmented development—clear context, reusable skills, and defined workflows that keep quality and consistency high while moving faster.