Skip to main content
The setup every other recipe assumes. It resolves which brand you are working as and caches it, so the pipeline and generate steps pick it up without re-asking.
1

Resolve the brand

list_brand_kits and pick the active one: the only one, the default, or ask the user which.
2

Read the voice

get_brand_kit and summarize the tone, key vocabulary, and any banned words back in a line.
3

Cache it

Write active_brand_kit_id (and any default avatar and connected accounts) to .contenthero/context.md at the workspace root.
Now the pipeline and caption recipes pick the brand up without re-asking. Revalidate a cached id with a live read before any spend or publish.