[BR|EN]~/samuel-felix

cat cases/este-site.md

This site

Bilingual, static, zero-cost portfolio with infrastructure as code and automated deploys.

Summary

This site is a project like the others: it has recorded decisions, a pipeline, infrastructure as code and stated limits. It’s all in the repository for you to check.

Home page weight 7.0 KB (2.8 KB compressed), with inlined CSS and no framework JavaScript
CI checks about 30 s (check, build, links and Terraform)
Cost US$ 0

Why it exists

Architecture

All page text lives in Markdown, in a content/ folder separate from the site code. That folder will be the chatbot’s knowledge base, with no copied text.

Decisions

Each decision has an ADR (architecture decision record, in Portuguese) with context, alternatives and consequences:

Pipeline

every PR and push
  npm ci                      dependencies locked by the lockfile
  astro check                 types and content (Markdown follows a schema)
  astro build                 generates the HTML
  lychee                      no broken links
  terraform fmt + validate    infra is still valid

push to main, all green
  pending-content gate        content marked as pending never reaches production
  wrangler pages deploy       publishes the build that passed to production

There are no preview URLs: staging runs on my machine with docker compose up, on the same Node version as CI and reachable only from localhost. PRs are for review and merge.

The pending-content gate exists because some numbers in the cases depend on material I haven’t received yet. Until then, they stay marked in the text and the production deploy refuses them.

Limits

Roadmap

Found along the way

Next: chatbot

A chat about the cases, answering from these same texts. The focus is on how it’s run and evaluated: evals in CI that grow with every real failure, abuse protection (invisible captcha, rate limiting, a daily usage cap), a kill switch and zero cost.

Further ahead

New cases.