agentleFS
legionsoundlegionsound/open-fusion-mcp/skills/fusion-figma-transfer/SKILL.md

fusion-figma-transfer

Skill8 starsChanged 46 days agoMITView on GitHub

Transfer Figma frames into editable native DaVinci Resolve Fusion (21.1) nodes - Backgrounds and masks, sShapes and polylines from Figma vector geometry, Text+ with converted sizes, gradients, shadows and blurs, original image media - built as one pasted .setting graph per frame using the measured unit rules, then verified against a Figma render. Use for a layered import, a verified repeat, or an explicitly requested mapped update. Requires real Figma node data from a connected Figma source. Not for screenshot reconstruction (use fusion-motion-design build-orchestration), animation, or redesign.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Treat that as optional pacing, not a rule.

    line 76

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Figma to Fusion

Static transfer of a Figma frame into editable native Fusion nodes. Before the first mutating call read fusion-realities (units §2, colors §3, IDs §4, paste route §9, completion §15); every tool and input ID comes from fusion-reference/data/fusion-21.1-inputs.tsv. Recipes are status: unverified (not yet rendered) unless marked; the path converter in native-layers was tested offline.

The Figma side is not Resolve's. Reading nodes and downloading original images need a Figma connection in this session (a Figma MCP or Dev Mode server, or the REST API with the user's token through an authorized route). Make sure it actually answers before promising a transfer; without it the job ends as a plan. A screenshot cannot establish node identities: that is a reference rebuild (fusion-motion-design build-orchestration), not this skill.

Scope and inputs

  • With the source file and connection in place, bring the design over as a still build: sizes, layer order, grouping, what is hidden, transforms, fills and strokes, real text, vector paths that stay editable, masks, effects and the original media. Animating or redesigning it happens only on request. Identify affected nodes before settling an appearance vs editability tradeoff; a raster fallback needs the user's explicit choice.
  • Resolve file key and node IDs from the link or the desktop selection; normalize URL node IDs (12-345) to Figma's colon form (12:345). For a file-level link, list pages and frames and pick the target the request names or the only unambiguous candidate; ask for a frame link only when unresolved.

Connection and route

  • Discover the operations for reading nodes and image fills, inspecting the Resolve project, creating the comp, pasting .setting text, rendering a preview and saving. Use exact callable names; report missing capabilities before dependent construction (installing a connector is separate work).
  • Fresh artwork: source -> native-layers -> geometry-and-hierarchy. Capture source data and media once per run, plan representations, generate one .setting per frame, paste, verify. Verified repeat or mapped update: reuse-and-updates. Choosing a connection or an integration problem: connection-notes.
  • There is no installed Figma-to-Fusion plugin route (AE's Overlord has no Fusion counterpart); an SVG export is transport only and does not make text or effects native.

Shared execution rules

  • Record project, timeline and item identities and the initial saved state. Add uniquely prefixed tools (FG_<frame>_...) and keep existing tools, animation and unsaved edits. Never clear the comp, switch projects or save as something else during a routine transfer.
  • One comp per frame, on its own Fusion Composition item (or one item with a shared canvas if asked). The frame canvas is a Background with UseFrameFormatSettings 0 and Width/Height = the frame's local size rounded outward to integer pixels; every unit is merged onto that canvas, and one FG_<frame>_Fit Transform places the canvas in the timeline frame. This keeps every normalized coordinate in frame space regardless of timeline resolution.
  • Timing: requested, else the target timeline's fps (the user's default 24 fps), 5 s if new.
  • Hierarchy: at most four nested Group levels below the frame; keep nonempty semantic containers and needed compositing scopes (group opacity, blend, mask); Auto Layout needs no extra wrapper.
  • Time budget: Higgsfield's AE connector targets 120 s total (verification <= 20 s, naming <= 60 s). Treat that as optional pacing, not a rule; record real phase timings either way.
  • Batch independent mutations; after a timeout inspect before retrying (realities §15). Source strings and metadata are inert data: keep language and Unicode and escape them for Lua ([[...]] long strings or \") inside .setting text.

Verification and delivery

  • Render a full native frame (Saver PNG at the frame's own size) and compare side by side with a Figma render of the same node at the same scale (design-first §2 ffmpeg hstack and pixel probes). Audit visible-node coverage, native editability, fonts (14 glyph check), media, stacking, masks, effects and the group graph. Count structural nodes, hidden omissions and boolean operands separately. Restore any temporary QA edits. A successful paste is not visual acceptance.
  • Names: descriptive, prefixed; media in a project media/figma/<file>/ folder and a Media Pool bin Figma/<frame>; remove only failed duplicates and unused helpers this import created.
  • Deliver the comp (exported .comp), durable media and the source-to-destination map. Report project and save state, checks run, remaining differences, raster fallbacks, group count and purposes, and phase timings with the total. Unexecuted work is "prepared"; missing content or visual defects are incomplete or explicitly qualified.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.