> ## Documentation Index > Fetch the complete documentation index at: https://docs.frontic.com/llms.txt > Use this file to discover all available pages before exploring further. # Initialize **Initialize** reads your live storefront once and writes the first draft of everything an agent needs to know about your shop — your visual language, your tone of voice, what you sell, how you sell it, and your brand as tokens. It's the fastest way to fill a Context Base, and it's what opens the first time you enter [Studio](/studio/frontic-studio) or the [Experience Designer](/designer/overview) on a fresh project. Initialize is in **Preview** and rolling out gradually. If you don't see it on a new project, you can build the same documents by hand — see [Rules](/context-base/rules) and [Skills](/context-base/skills). The problem it solves is the blank page. A Context Base is worth a lot once it exists, and writing one from scratch is work nobody wants to do before they've seen the product do anything. Initialize does the first pass for you and leaves you editing rather than authoring. ## Running it You'll usually meet Initialize on your way somewhere else: open Studio or the Designer on a project that hasn't run it, and Frontic steers you here once. **Skip for now** is always available and is remembered — you won't be asked again. You can also start it deliberately from **Library → Brand** on a project with no brand yet. Enter your storefront address. If your project already has a Shopware integration, Initialize reads catalog data through it as well; otherwise you can pick another integration or let it read the storefront only. Takes about 5–10 minutes. You can leave and come back — the reading continues in the background, and a banner across Studio and the Designer tells you where it's up to. Your storefront answers most of it. What's left comes as a short interview — usually a couple of minutes. Initialize writes its documents and hands you into the Designer. Everything it wrote is editable from that moment on. ## What it reads The reading walks your shop the way a shopper would, then the way an analyst would. | Step | What it looks at | | ------------------------ | ---------------------------------------------------------------------------- | | **Reading your shop** | Your homepage and the links customers see first — navigation, header, footer | | **Reading your theme** | Your stylesheets, and how the config compares to what shoppers actually see | | **How you speak** | Tone of voice across your pages, down to formal versus informal address | | **Your visual language** | Every color your CSS ships, and which one carries the brand | | **What you sell** | Assortment, categories, manufacturers, filters, where the value sits | | **How you sell** | Campaigns, offers, discounting, and the promises your shop makes | As it goes, it shows you what it found — the colors it weighed, the link map, the tone markers, the manufacturer mix, the numbers behind your catalog. When your config and your live storefront disagree, it sides with the storefront. ## The interview Some things a storefront can't tell you: who you're really selling to, where you stand against competitors, which of two plausible readings is the right one. Initialize collects exactly those and asks them one at a time, grouped by topic and marked by impact. Every question has an escape hatch: * **You decide for me** — Initialize picks from your storefront plus best practice * **I'll tell you** — answer in your own words instead of choosing an option * **Not relevant** — skip it The count of settled decisions ticks up as you go. When the storefront answered everything, the interview is skipped entirely. ## What it writes Initialize produces eight documents and puts each where the agents that need it will find it. | Document | Lands in | Read by | | ---------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------- | | **Styleguide** | [Context Base rules](/context-base/rules) | Every agent, when the task touches visuals | | **Tone of voice** | Context Base rules | Every agent, when it writes copy | | **Product domain** | Context Base rules | Every agent, when the task touches the catalog | | **Marketing playbook** | Context Base rules | Every agent, on campaign work | | **Store setup guide** | [Context Base guides](/context-base/guides) | Pulled in on demand | | **Project rules** | Studio's [Project Rules](/studio/settings#project-rules) *and* the Designer's [Project Rules](/designer/building#designer-settings) | Always applied, on both surfaces | | **Designer prompt** | The Designer's [Designer Prompt](/designer/building#designer-settings) | Every Designer session | | **Brand tokens** | The [Design Library](/context-base/design-library) | Every experience the Designer builds | Project rules land in **two** places on purpose — Studio and the Designer each get their own copy, so you can let them diverge as the projects' needs pull apart without breaking the other one. Nothing here is final. These are drafts written from evidence, and the whole point is that you now have something concrete to correct rather than a blank editor. Read the Styleguide and Tone of voice rules first. They're the two that shape the most output, and they're the two where a small correction — a color that isn't really your brand color, a "we address customers informally" that should be the other way round — pays off across every future session. ## Running it again Re-running Initialize overwrites the documents it owns with a fresh reading. Useful after a redesign, a repositioning, or a catalog shift big enough that the old drafts are wrong. Brand tokens are handled more carefully: a re-run doesn't touch your live brand while it's reading. The new tokens only replace the old ones once you've been through the interview and confirmed, so a re-run can't leave experiences mid-build without a brand. Earlier brand versions stay in the Design Library's history either way. ## When it can't read your shop Initialize is a real browser hitting a real storefront, so it runs into the same walls anyone else would. It tells you which one it hit. | Problem | What to do | | ---------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Address unreachable** | Check the spelling and that the shop is online | | **Shop returned an error** | It may be down or password-protected | | **Timed out** | Try again in a moment | | **Blocked by your host** | Ask your host to allow Frontic | | **Blocked by your firewall or bot protection** | Add the shop's domain to the project, create a bypass token under **Settings → Secrets**, then allow requests carrying the `Frontic-Bypass-Token` header in your WAF | A failed reading changes nothing — your shop isn't touched, and you can start it again whenever you like. Frontic only sends your bypass token to domains registered on the project. That's why the domain has to exist before the token is any use — see [Domains](/project-settings/domains). ## Related The brand tokens Initialize derives, and how to refine them. Where the styleguide, tone of voice, and domain rules land. What the Context Base is and how agents read from it. The surface Initialize hands you into when it's done.