This space documents the systems behind machinebehavior.io, tychat.io and uncovertechtalent.com: how each site is built, the conformity gate that every push passes before it serves readers, the deploy job, the crawler that draws the map of the work, the Inside front page and this documentation tree.
Three static sites, one gate. Each site is static files on GitHub Pages. Each deploy runs the same conformity checks first; a failed check stops the deploy and the previous build stays live.
Systems
| System | What it does | Source | Page |
|---|---|---|---|
| machinebehavior.io | Research site: claims, experiments, case files, Inside, the map, these docs | uncovertechtalent/machinebehavior.io | Site architecture |
| Conformity gate | Checks every page on every push, weekly and by hand; blocks the deploy on a failed check | .github/actions/conformity/run.py | Conformity gate |
| Deploy job | Refreshes the map, uploads the site, deploys to Pages | .github/workflows/conformity.yml | Deploy pipeline |
| Map crawler | Crawls the three sites and Substack into one graph | scripts/crawl_map.py | Map crawler |
| Inside | Front page with live status, deploys, experiments and dashboards | inside/index.html | Inside portal |
| Docs tree | This documentation, from markdown and the knowledge vault | scripts/build_docs.py | Docs tree |
| Observability | Prometheus, Loki and Grafana behind the public dashboards | agent-observability | Observability space |
Where to start
- New to the setup: read Site architecture, then Deploy pipeline.
- Adding a page: follow Add a page to machinebehavior.io.
- A deploy did not go out: start at Runbooks.
- Why something is the way it is: the Decision log.