Walkthrough
Why single-file HTML is a delivery format, not an authoring format
A walkthrough of the architecture this skill recommends, told through the same blocks the skill provides.
The temptation, and the cost
Generative models are happy to emit one giant HTML file on request. The result looks impressive in isolation. But the moment you want to change something, the cost of that approach lands.
The shape of a working pipeline
The architecture this skill recommends has four moving parts: source content, a compiler, a normal website, and a packager that runs once at the end.
How big is "single-file" really?
Inline assets dominate the file size. The biggest cost is Base64 — about a third on top of the raw binary. Try the controls below to feel the relationship.
Bootstrapping a project
A new project is two dependencies plus the single-file packager. Choose the package manager that matches your team.
A worked example: tracing an incident in one HTML file
A small incident report fits comfortably as one of these deliverables. Timeline plus annotated diff plus follow-up actions.
What to take from this
A finished single-file HTML deliverable should feel like a designed product. The way you get there is by treating it as the last step in a normal website pipeline, not as the authoring artefact.