Tools

Plumb exposes twenty MCP tools, split across the two directions it plumbs — Figma → code (read) and prompt → design (write). Each one has a focused, single responsibility; the agent composes them.

Read — Figma → code

Extract a design as a compact spec, build it, then diff the render against the design and self-heal to a pixel-perfect match.

Tool	What it does
plumb_status	Self-description, key legend, connection state. Call first.
plumb_outline	Every screen in the file (id, name, size).
plumb_node	Extract a screen as compact PDS — by id or by name.
plumb_query	Pull a slice of a screen by pattern (skeleton / buttons / text / components) when the full tree would blow the token budget.
plumb_describe	Text-only visual description — per-region narrative + child summary, for image-blind harnesses or token-conscious flows.
plumb_tokens	Design-token table (colours, type, radii, shadows).
plumb_selection	The user's live Figma selection.
plumb_assets	Export icons (SVG) + images (PNG) — three modes.
plumb_screenshot	Render any node to PNG/JPG.
plumb_search	Find nodes by name and/or type.
plumb_components	List components + instance usages.
plumb_verify	Diff your rendered layout against the design — ΔE2000 colour distance, shadow/rotation/flex-child/fill-stack checks.
plumb_fit	The self-healing loop: plumb_verify plus a 0–100 convergence score and prioritised fixes, so the agent iterates to pixel-perfect instead of one-shot checking.
plumb_fig_outline	Headless: list every screen in a saved .fig file from disk.
plumb_fig_node	Headless: fetch one node from a saved .fig file by id.

Write — prompt → design (the director)

Turn a one-line brief into a full, on-brand Figma design: research real reference sites, extract a brand, compose the page from real nodes, then critique the render and iterate — with no extra API key, because the agent driving the MCP server is the creative director.

Tool	What it does
plumb_studio	The design director. One brief → researched references → extracted brand → a full composed Figma page (nav, hero, features, gallery, CTA, footer). Returns the node ids + authored spec so you can critique and refine.
plumb_brand	Brief → live-screenshots best-in-class reference sites + a synthesized brand palette / type board (real Figma Variables + text styles) on the canvas.
plumb_design	Author a design from Plumb's high-level Design DSL and build it into Figma — full control over pages, sections, blocks, and brand tokens. The refine step of the loop (mode:"sync").
plumb_review	The critique loop — the mirror of plumb_fit. Blends a structural diff, a deterministic design rubric, and the calling agent's own vision verdict into one score + ranked fixes.
plumb_source	Resolve on-brief assets (icons, photos, avatars, illustrations, patterns) for a design — ranked candidates or a downloaded best match.

Tool selection

If you only remember three per direction:

Read (design → code):

• plumb_outline — figure out which screens exist.
• plumb_node — get the design spec for one of them.
• plumb_fit — build it, then iterate to a pixel-perfect match (or plumb_verify for a one-shot check).

Write (prompt → design):

• plumb_studio — a brief becomes a full, on-brand Figma page in one call.
• plumb_review — screenshot the build, grade it yourself, and get a ranked fix list.
• plumb_design — apply those fixes with mode:"sync" and loop until done.

The rest are how the agent fills in the details — exports, screenshots, search, design tokens, component awareness, sliced queries for dense screens, text descriptions for image-blind harnesses, brand research, and asset sourcing.

How the agent picks paths

Most read tools accept either id (canonical) or name (looked up against the live outline). Pass id whenever you have it; it's unambiguous. Names can be duplicated — Plumb handles that by returning a list of matches for the agent to disambiguate.

Read tools that need design data auto-pick between the plugin path (instant, no rate limits, requires Figma open) and the REST path (headless, rate-limited, requires FIGMA_TOKEN). With the plugin paired, omit fileKey. For REST, pass fileKey + id.

The plugin-path tools — plumb_outline, plumb_selection, plumb_assets, plumb_screenshot, plumb_search, plumb_components — require the plugin paired and won't fall back to REST. The other read tools (plumb_node, plumb_query, plumb_describe, plumb_tokens, plumb_verify, plumb_fit) work on both paths.

Every write tool builds through the plugin path — plumb_studio, plumb_brand, and plumb_design all require the Plumb plugin paired to write to Figma (plumb_design can dryRun to compile + validate without it). plumb_review is read-only — it grades the emitted result — and plumb_source needs no Figma connection at all.