Figma merges code, context and agents on one canvas

Figma has announced it is bringing code, context, and agents together on what it calls the full-stack creative canvas. The move signals that the company now positions itself not as a drawing tool but as the environment where interface work and implementation meet.

The announcement, reported by CXOToday.com, names three ingredients: code, context, and agents. Each one changes a different part of the designer's workflow, and together they describe a canvas that carries a design from first frame to running product.

What does "code on the canvas" change for practitioners?

For most of its history, Figma treated code as an export. Designers composed frames; engineers rebuilt those frames in an editor somewhere else. The handoff document, the redlines, and the named layer conventions were all workarounds for that separation.

Placing code inside the canvas collapses that boundary. The file stops being a picture of an interface and becomes a working artifact. A practitioner can move from visual adjustment to the logic behind it without switching tools.

Why does context matter as a first-class material?

Context is the second pillar. Design systems already carry component libraries and variables; context extends the idea to the information a project accumulates over time — decisions, constraints, and prior versions.

The practical consequence is reuse with judgment. When a canvas understands its own context, a designer who opens a file inherits the reasoning behind it, not just its geometry.

What do agents add?

Agents are the third element, and the one that restructures the work itself. An agent that operates on a shared canvas can act on the same objects a human edits. Instead of prompting in a separate window and pasting results back, the practitioner delegates work inside the file where the record of it stays.

Framed as a "full-stack creative canvas," the announcement also claims a wider scope: not screens alone, but the whole stack a product runs on.

Who should watch this closely?

Three groups face near-term changes:

• Interface designers, whose files now border on production code and whose review responsibilities may grow accordingly.
• Front-end engineers, who receive artifacts closer to what they ship and may lose part of the translation layer they own today.
• Design-system leads, whose component definitions gain a second runtime — the agent — as a consumer alongside human teams.

The comparison worth holding onto: this is the same structural shift that happened when page-layout tools absorbed prepress controls. The work did not disappear; it moved upstream into the design file, and the people who understood both sides of the boundary became the scarce ones.

Figma has not yet published full technical documentation of the three capabilities, so questions remain about versioning, permissions, and how agents will operate across team libraries.

How quickly teams adopt the full-stack canvas will depend less on the demo and more on whether the code it produces survives contact with real engineering standards.