Breno Brito website

Source for https://brenorb.com.

This repo contains Breno Brito's personal site, including:

• articles and notes in _posts/
• profile and static pages such as /about/, /projects/, and /media/
• machine-readable crawler context in llms.txt and llms-ctx-full.txt

Requirements:

• Ruby with Bundler
• the gems from Gemfile

Install dependencies:

bundle install

Run locally:

bundle exec jekyll serve

Build the site:

bundle exec jekyll build

Verify that the Search Console regression routes exist and point to their canonical pages:

ruby scripts/check_indexing_routes.rb

Install the repo hooks:

./scripts/install-git-hooks.sh

The pre-commit hook and CI both enforce a size limit for local feature: images in posts:

ruby scripts/check_feature_images.rb

They also verify that embedded media posts are covered by the transcript manifest and wired to generated chapter includes:

node scripts/check_media_sources.js

Audit post metadata coverage:

ruby scripts/audit_post_metadata.rb

Generate diarized raw transcripts one item at a time:

node scripts/transcribe_media_batch.js

Regenerate the per-post appendix data used by the site:

node scripts/build_media_appendices.js

That appendix build keeps the raw transcript JSON untouched and only cleans obviously pathological repeated-word runs in the display layer, so lines like we we we we render as we... we.

Verify that every post with an embedded player is covered by the transcript manifest:

node scripts/check_media_sources.js

Verify that every manifest item has raw JSON, native --script output, and appendix data:

node scripts/verify_media_transcripts.js

Regenerate the chapter includes for posts that already have curated chapter specs:

node scripts/build_topic_chapters.js

• _config.yml holds site metadata, SEO fields, and Jekyll settings.
• scripts/audit_post_metadata.rb reports missing post metadata such as description, excerpt, tags, and feature.
• scripts/build-llms-ctx-full.rb regenerates the expanded AI context file.
• scripts/media_sources.json is the media manifest for transcript generation.
• scripts/check_media_sources.js verifies that embedded media posts are present in the manifest and include generated chapter navigation.
• scripts/topic_chapters.json is the curated chapter source of truth for media posts.
• scripts/build_topic_chapters.js renders those curated chapters into _includes/generated/.
• scripts/verify_media_transcripts.js verifies that each manifest item has raw transcripts, native script output, generated appendix data, and published diarization metadata.
• scripts/transcribe_media_batch.js runs the frozen fscript + diarization flow sequentially and writes both JSON and native --script outputs.
• scripts/build_media_appendices.js converts raw transcript JSON into _data/media_appendices/.
• control.sh provides a small Podman-based workflow for containerized local serving.

For new posts, use this minimum front matter baseline:

• title
• excerpt
• description
• tags
• feature for shareable pages and posts

Use content_type when the post should render outside the default writing archive, especially for media and project entries.