figcraft

FigCraft — headless UI design

FigCraft is ArtCraft's interface-design app (pages, frames, auto layout, components, variables, prototyping; 1 100+ commands). pi-craft provisions the headless figcraft-cli on first use (no desktop app). App choice, the discover → execute → save → verify loop and the MCP policy live in craft-studio.

What it is for / formats

Screens and design systems. Native file: .figcraft (JSON). Opens .figcraft .json (FigCraft or REST API file JSON) .fig (best effort; info lists what was not imported) .svg and rasters (.png .jpg .webp .gif .bmp → an image layer). Exports png jpg webp svg pdf and .fig (experimental). For free-form vector illustration use VectorCraft; for page layout/print DesignCraft.

CLI dialect

pi-craft figcraft exec <args…> passes the arguments to figcraft-cli. There is no --help (use help); --no-system-fonts on any subcommand makes text render identically on every machine.

figcraft-cli commands [--json] [--filter TEXT]        # text listing ("id*  label" + params line) or JSON; no describe command
figcraft-cli run SCRIPT [--in FILE] [--out FILE]      # SCRIPT = .jsonl (one step per line) or .json (array) of {"command": id, "params": {…}}
figcraft-cli info FILE [--json]                       # pages, frames with ids, node count
figcraft-cli export FILE -o OUT [--format png|jpg|svg|webp|pdf|fig] [--scale N] [--page NAME] [--node ID|NAME]
figcraft-cli render FILE -o OUT.png [--scale N] [--page NAME]
figcraft-cli convert IN OUT                           # e.g. svg -> .figcraft, .figcraft -> .fig

• Scripts are files, not --cmd flags: write the JSON lines to a file and run it. The result of each step is a JSON line ({"ok":true,"result":{"id":2},"step":0}) — ids are small integers assigned in creation order (the first frame is 2), so later steps reference earlier ones with "parent": 2.
• export -o OUT: when OUT has an image extension (or --node is given) one file is written; otherwise OUT is a directory that gets one file per top-level frame named after the frame (--format picks the type). A PDF of a file has one page per top-level frame.
• --node takes an id or a layer name (--node Button); --scale 2 doubles pixel size (a 393×852 frame → 786×1704). render is the same as a png export without --node.
• In-script export: file.export {ids, format, size: "2x" | scale, path} writes during the run (relative to cwd).