Cross-pollinate extraction with brand context

A website is only one expression of a brand, so extracting it in isolation misses intent that lives elsewhere. In real engagements you want to cross-pollinate the extraction with the surrounding context. Anything the editor can read works: another app, a design system exported from Figma via MCP, a brand guide PDF, a screenshot or poster, an existing DESIGN.md, an existing token or theme file (tailwind.config, CSS custom properties), a Storybook or component library, a DTCG token JSON exported from Tokens Studio, or a competitor extraction to benchmark against. In an agentic workflow this is trivial. Open Claude Code, Copilot, or your preferred editor, drag and drop the supporting files into context, and let the agent crawl the target with Dembrandt alongside them. Dembrandt's native Playwright extraction supplies the measured, real-world tokens; the loaded documents supply intent and ground truth. Together they produce a reliable starting set of design tokens and a clear map of where the live product diverges from the brand it is supposed to express.