plumb_studio_start

Step 1 of 3 in the transparent studio flow — the same director pass as plumb_studio, split into watchable steps across separate named Figma pages instead of one opaque call. Opens a session that plumb_studio_kit and plumb_studio_page reuse.

How it works

Given a one-line brief, this step does everything plumb_studio does except compose the final page:

brief → research references → screenshot them live → extract a brand → build a References + Brand board → open a session

1. Researches references — discovers best-in-class sites for the brief (plus any explicit references, always studied first).
2. Screenshots them live — renders each in headless Chrome, reused later as gallery/hero imagery by plumb_studio_page.
3. Extracts a brand — synthesizes a coherent semantic palette + type scale from the captures.
4. Builds a References + Brand board — real Figma Variables + text styles, written onto its own page (default "Brand").
5. Opens a session — returns a sessionId that carries the brand, references, and already-staged screenshots forward, so plumb_studio_kit/plumb_studio_page never re-capture anything.

Requires the Plumb plugin to be paired — this tool writes through the plugin path.

Inputs

Field	Type	Notes
brief	string · required	One-line description, e.g. "a premium fintech dashboard". Drives reference discovery, palette, and copy for later steps.
count	number · optional	How many reference sites to study + screenshot. Default 4.
references	string[] · optional	Explicit reference URLs to include first — always studied ahead of the discovered catalogue.
pageName	string · optional	Figma page for the brand board. Default "Brand".

Returns

{
  "sessionId": "sess_abc123",
  "name": "Fintech Dashboard",
  "brand": { /* synthesized semantic palette + type scale */ },
  "references": [{ "name": "Mercury", "url": "https://mercury.com" }],
  "captured": 4,
  "misses": [],
  "foundationsWarnings": [],
  "rootId": "412:9",
  "next": "Watch the \"Brand\" page. Then call plumb_studio_kit(\"sess_abc123\") to build the component library."
}

Example

"Use plumb_studio_start to direct a brand for a premium fintech dashboard.
 Watch the Brand page, then call plumb_studio_kit with the returned sessionId."

When the agent should use it

• You (or the user) want to watch each stage build on its own Figma page instead of getting one finished page back from plumb_studio.
• You want to review or adjust the brand before any product page gets composed against it.

Notes

Follow with plumb_studio_kit, then plumb_studio_page (repeatable, once per page). For the one-call equivalent, use plumb_studio instead.