Community publishing is now open for Weave tools. Since launching 30+ custom-built Weave tools at Config 2026, Figma designer advocate Moran Michelle Dankner has been building tools for her own workflows — and anyone working in Weave can now publish theirs to the Figma Community, usable directly on the Figma Design canvas.

Sharing tools with the community

The tools launched at Config 2026 exist to make repeatable image generation tasks — producing on-brand icons, transferring a style reference to a new image — runnable inside Figma. Dankner's own releases, such as Professional Headshot, Blue Embroidery and Brand Stamp, began as personal experiments to test how far a tool can stretch.

Anyone with a paid Weave plan can publish tools to the Figma Community, and anyone with a full Figma seat can run them in Figma Design. Weave tools in Figma Design will consume Figma AI credits once generally available.

Five published tools worth trying

Interior Design Moodboard to Render — Mallory Dean

Reads the colors, materials, furniture and lighting in an interior moodboard and turns them into a photorealistic 3D room render. Built-in room types and camera angles guide composition. Dean's advice for tool builders: lean on an LLM for both writing prompts and structuring nodes, and feed results back as feedback during revisions.

Physical Architectural Model Maker — Yordan Vakarelov

Generates an architectural model from a reference photo, with custom materials for the building, landscape and finer details. Vakarelov's stated approach is to absorb the technical setup so users can stay on creative ideas without getting bogged down in prompting or mechanics.

Image Reframer — Jan Gericke

Extends an image to any target aspect ratio while matching the original lighting, shadows and perspective. Users can specify where the subject should sit in the new frame — for example, keeping the left quarter free for a headline or logo — so the extension produces clean negative space rather than a guess.

Sketch Generator — Eric Satoru

Turns a short text prompt into a hand-drawn concept sketch for an architectural project in the "what if" stage, before rendering or modeling. Prompts work best when they name the space, its materials and a view. Satoru's rationale: a finished-looking image closes the conversation, while rough sketches at the speed of a conversation let stakeholders react to ideas.

Nail Art — Stephanie Zhou

Personalizes a nail set and previews it on your own hands before committing to the real thing. Zhou built the harder, non-negotiable features first, such as 3D decorations and styles, then added dropdowns for nail length, shape, style and decorations.

Before you publish

A tool that returns a consistent output from any input depends on a chain of decisions made under the hood — and publishing means making those calls for everyone who runs it. Three considerations:

  • Decide who you're building for. Answer "who is this actually for?" before writing a single prompt; the audience changes what the tool unlocks and exposes, and even how inputs are labeled.
  • Design for unpredictable input. Handle both clean and messy submissions. Dropdowns and pre-made choices give users vocabulary while protecting the back end; an LLM layer between the request and the generation model can translate creative intent into something the model can work with.
  • QA as if you won't be in the room. Hand the tool to someone who didn't build it, give zero explanation, and watch where they pause and what they expect that never happens. Fix that before publishing.

Weave's help center covers how to build your own tool.