Breakgraph (breakgraph.com)

Made battle participant names clickable by converting them to search links, allowing users to quickly search for a participant when viewing a battle.

This PR introduces automatic confirmation of high-confidence battle video extractions at ingest time, eliminating manual review for videos that pass structural validation gates. The single-URL ingest page was firing all videos in one server action call. --- _Generated by [Claude Code](https://claude.ai/code/session_01WC3hJyMr5vwN3DyCHXLFrr)_. Netlify's CDN was caching /og responses by path only, ignoring query params, so all requests got the first cached image. Claude/add analytics seo k lsz e. Env vars needed in Netlify web site:. The /ingest rewrite proxy caused 400 errors from PostHog's backend, likely due to Netlify's edge network modifying the proxied request body.

Battle pages are cleaner — event scale indicators and winner confidence badges have been removed, since location and results already tell that story without the extra noise. Breadcrumbs on multi-year events no longer repeat the year.

You can now search for battles directly from the navigation bar, with a typeahead dropdown surfacing suggestions as you type. It's the fastest way to get to a specific battle or event without browsing through lists.

Direct profile pages for breakers and crews have been removed. The homepage has been restructured so visitors explore the breaking scene through battles and events instead — a more natural entry point. Entities now appear in context, where they're most meaningful.

Recurring events — like annual conferences held under the same name — were collapsing into a single entry instead of tracking each edition separately. Each event now gets a year-specific slug, keeping editions properly distinct. Also fixed several bugs causing duplicate records to accumulate across events, people, and results.

Three bugs were silently preventing key data from being saved — person roles, scene associations, and match results were all failing to write to the database. The data now saves correctly, so breakdowns of who appeared where and in what capacity are more complete.

The homepage header lost its descriptive subtitle, and the video detail page no longer shows the video description block. Both pages feel tighter.

The site header now includes links to the main content sections — Breakers, Crews, Events, Scenes, and Videos. On desktop they appear as a second row beneath the site name, with the current section underlined; on mobile the layout adapts to fit smaller screens.

Breakgraph now has a dedicated About page that explains what the project is and shows live stats on how much of the breaking community it currently covers. It's a useful first stop for anyone new to the site wondering what they're looking at.

Videos in Breakgraph now have their own pages, each with an embedded YouTube player so you can watch without leaving the site. It's a first step toward making Breakgraph feel like a proper discovery layer, not just a dataset.

Videos imported through the data pipeline were getting stuck in a holding status and not surfacing in the app. The fix ensures imported videos move into the active pipeline correctly, so new content shows up as expected after an import run.

Got the initial interface up — a knowledge graph for the breaking dance world. Early stage, but enough to test whether connecting people, crews, events, and music into a single visual graph makes sense.

The AI extraction pipeline was missing some crew members who appeared in the people data but weren't being linked as distinct entities. Fixed by deriving those connections from existing records rather than relying solely on direct extraction.