Figma Specs Turns Your Markdown Notes Into a Living Design Document

Figma just shipped Specs, a new document type that sits directly on your Figma or FigJam canvas and connects your written decisions to your designs, diagrams, and coding agents. If you have ever lost a key decision inside a stale Notion page or a local Markdown file nobody else can find, this one is for you.

The Problem Specs Is Actually Solving

Here is the thing about Markdown: it has quietly become the default language for talking to AI agents. When you prompt a coding tool or ask an AI to help you plan a feature, you are essentially writing structured text. Markdown is that structure. But a Markdown file sitting on your laptop is isolated. It has no live connection to your designs, no built-in comments, no way for a teammate or an agent to work on it alongside you.

Figma's argument with Specs is that some of that written context is not throwaway. It is the reasoning behind your decisions, the edge cases you mapped out, the user flow you agreed on. That context deserves to be a first-class object in your workspace, not a screenshot pasted into a doc that nobody updates.

What Specs Actually Does

A Spec in Figma is a multiplayer document container. You press D on the keyboard or click the Specs icon in the toolbar, and a rich text document opens right on the canvas, next to the designs it describes.

A few things make it different from a regular doc:

Embedded canvas objects. You can drag a FigJam diagram, a design frame, or any node from the Figma canvas directly into a Spec. The design and the words describing it sit side by side, not in a screenshot that goes stale the moment someone tweaks a component.

Drag-and-drop Markdown import. If you already have a .md file with notes or a rough PRD, you drag it onto the canvas and it becomes a Spec instantly. No copy-pasting.

Agent-readable comments. The Figma agent can read and reply to comments on a Spec, the same way a human collaborator would. That means you can ask the agent a question inline and get an answer anchored to the exact line of text where the question lives.

Connectors for outside context. A GitHub connector lets you pull in context from a codebase when drafting a technical spec. A Hex connector (Hex is a data analysis tool) lets you ask the agent to run a product usage analysis and drop its findings as a comment right where you posed the question.

How a Designer Building With AI Would Actually Use This

Say you are vibe-coding a simple feature, maybe a bookmarking flow in a mobile app. Here is a realistic workflow with Specs:

1. Draft your intent. Press D on the canvas near your mockups and write out what the feature should do, who it is for, and any edge cases you have already thought through. Keep it rough. This is your thinking, not a final doc.

2. Embed your designs. Drag the relevant frames from your canvas into the Spec. Now the written intent and the visual exploration are one object. A teammate reviewing the Spec sees both without switching tabs.

3. Invite the agent in. Ask the Figma agent to fill gaps, suggest edge cases you missed, or draft a more structured version of your notes. Because the agent can read comments, you can have a back-and-forth inside the document itself.

4. Send it to your coding agent via MCP. MCP (short for Model Context Protocol) is a standard that lets AI coding tools read context from Figma. Once your Spec is solid, you share it through Figma's MCP server so your coding agent starts with your actual intent, not a guess. The article describes how a Figma engineer used exactly this loop when building the Specs feature itself.

5. Keep the Spec alive. As the build progresses, update the Spec on the canvas. Ask agents to generate new diagrams in FigJam, embed those, and share them with whoever is doing code review. The document grows with the work instead of going stale.

What to Watch For

Specs looks genuinely useful for solo designers starting to build, but a few honest questions are worth keeping in mind.

The connectors (GitHub, Hex, and others from a catalog of 20 new integrations) add real power, but they also assume your team is set up to use them. If you are a solo designer experimenting with AI-assisted building, you may not have a codebase on GitHub or a Hex account yet. The core Spec workflow, writing, embedding designs, collaborating with the agent, still works without any connectors.

It is also worth noting that Specs lives inside Figma. If your team's written work lives primarily in Notion, Linear, or Confluence, there will still be some friction. Specs does not replace those tools yet; it adds a canvas-native layer for the decisions closest to your designs.

That said, for designers who are already in Figma all day and are starting to use AI to turn those designs into real code, Specs closes a gap that has been quietly annoying for a while. Your reasoning no longer has to live somewhere else.