Back to list
Oct 16, 2026
45
0
0
AI ToolsNEW

Figma Specs: Markdown Spec Docs on the Canvas for Agents

Figma Specs puts Markdown specs on the canvas with GitHub and Hex connectors, agent comments and MCP handoff. Open beta on paid plans from Oct 15.

#Figma#Figma Specs#Markdown#AI Agents#MCP
Figma Specs: Markdown Spec Docs on the Canvas for Agents
AI Summary

Figma Specs puts Markdown specs on the canvas with GitHub and Hex connectors, agent comments and MCP handoff. Open beta on paid plans from Oct 15.

Introduction

On October 15, 2026, Figma introduced Figma Specs, announced at its first Config event in India according to TechCrunch. Figma's blog describes a spec as a multiplayer container for Markdown context that moves into, out of and across Figma. The idea is to give teams, and the coding agents working with them, one shared document that sits next to the design work it describes instead of living in a separate wiki.

Figma's chief design officer Loredana Crisan told TechCrunch: "17-times more lines of code being produced today than in the past. But when you're looking at releases, only 1.3-times more releases." Specs is positioned as a way to define what gets built before code is generated.

Feature Overview

Markdown on the canvas. You can drag any .md file onto the canvas to turn it into a spec, or create one by pressing D or using the Specs icon in the toolbar in Figma or FigJam. A spec can embed a FigJam file, a frame or any canvas node, plus diagrams, code layers and designs. Specs can be duplicated so teammates work on separate versions before aligning, and they can be synthesized from feedback such as annotations, stickies and comments.

Connectors. A GitHub connector can draft a technical spec from your codebase. A Hex connector can run an analysis on product usage data and leave the findings as a comment next to the related question in the spec.

Figma agent. The agent can read and write comments in a spec, for example replying to every open comment with a suggested fix, each in its own thread. It can read a spec as Markdown and edit it, such as updating a feature name across the document. Figma says it can turn a PRD into designs, prototypes and code, or turn designs and code back into a spec.

Code layers. Code layers moved into open beta the same day. Their "Extract designs" action converts working code back into editable design layers, and you choose whether to bring over a single screen, a specific state or a full flow.

Agent handoff and MCP. To hand work to a coding agent, copy the spec link into the agent's prompt. Through the Figma MCP server, agents such as Codex can read the spec and the context it links to: Markdown, frame references, diagrams, code layers and Code Connect snippets. Figma's update says MCP now has reduced errors and token usage, lets agents generate images, create code layers and run Weave workflows, and adds 20 new integrations. TechCrunch names Open Code, Dia, Grok Bot and Sarvam among them.

Usability Analysis

Figma's own examples are modest and concrete. Mark Harvey, Staff Product Designer at Intuit, uses specs beside icons to document file names, categories, tags and descriptions, saying a consistent structured document makes design files easier to follow for non-designers. Other Figma staff describe consolidating brainstorm decisions into a spec and using a spec as the context given to an agent through MCP.

The strength of the design is placement. Because the spec lives on the canvas, designs and diagrams can be dragged into a spec and back out with their Figma properties preserved, and comments from people and the agent share one thread model. Specs works on web, desktop and mobile.

The workflow assumes a team already organizes work in Figma and uses a coding agent that connects through MCP. Teams that keep requirements in other tools will need to decide whether moving them is worth it, though the drag-and-drop .md import lowers that cost.

Pros and Cons

Advantages include a single place for requirements and design references, Markdown that stays portable, agent participation through ordinary comments, and a link-based handoff to coding agents. The GitHub and Hex connectors add codebase and usage-data context without leaving the document.

Limits come from the rollout. Specs is in open beta for paid plans starting October 15, so free-plan users are not covered at launch. Figma gave no pricing for Specs in the sources reviewed. The value also depends on agent output quality, which Figma says its steering tools aim to make more predictable. Finally, the surrounding features, including code layers, are still in beta.

Outlook

Crisan's comparison of code volume against releases frames the product thesis: the bottleneck is deciding what to ship, not producing code. If Specs becomes the shared reference that agents read through MCP, it positions Figma as a context layer for agentic development rather than only a design editor. TechCrunch also reports that Figma is connecting Payload, the open-source headless CMS it acquired, to Figma Make, with Payload downloads up sixfold over 12 months, which hints at a broader push from design toward shipped websites.

Conclusion

Figma Specs is a focused addition: Markdown documents with canvas references, agent comments and a clean handoff path to coding agents. It suits product and design teams already on paid Figma plans that are experimenting with agent-assisted development. Teams outside the paid tiers, or without MCP-connected agents, will get less from it today. Rating: 4 out of 5.

Editor's Verdict

Figma Specs: Markdown Spec Docs on the Canvas for Agents earns a solid recommendation within the AI tools space.

The strongest case for paying attention: a shared Markdown document on the canvas gives teams one reference for requirements and design context. That alone raises the bar for what readers should expect in this space. Reinforcing that, agent replies arrive as separate comment threads, so every suggestion can be reviewed or ignored — practical value rather than just headline appeal. The broader signal worth registering is straightforward: the launch rests on the view that deciding what to build, not generating code, is the bottleneck; Figma's design chief cited 17x more code but only 1.3x more releases. On the other side of the ledger, one constraint is real rather than a marketing footnote: the open beta covers paid plans only, so free-plan users have no access at launch. It should factor into any serious decision. Layered on top of that, related features such as code layers are still in beta, so behaviour may change — which narrows the set of teams for whom this is an obvious yes.

For product, design and engineering teams already on paid Figma plans and experimenting with coding agents, this is a serious evaluation candidate, not just a curiosity to bookmark. For everyone else, the safer posture is to monitor coverage and revisit once the use cases that matter to your team are demonstrated in the wild.

Advertisement

Pros

  • A shared Markdown document on the canvas gives teams one reference for requirements and design context.
  • Agent replies arrive as separate comment threads, so every suggestion can be reviewed or ignored.
  • The GitHub and Hex connectors bring codebase and usage-data context into the spec.
  • A copied spec link is enough to hand context to a coding agent through MCP.

Cons

  • The open beta covers paid plans only, so free-plan users have no access at launch.
  • Related features such as code layers are still in beta, so behaviour may change.
  • The workflow assumes a coding agent connected through the Figma MCP server, which not every team uses.
  • Output quality still depends on the agent, which Figma says its steering tools aim to make more predictable.
Advertisement

Comments0

Key Features

1. Markdown specs: drag a .md file onto the canvas or press D to create one in Figma or FigJam. 2. Embeds: FigJam files, frames, canvas nodes, diagrams, code layers and designs. 3. Connectors: GitHub drafts a technical spec from your codebase; Hex leaves analysis findings as comments. 4. Figma agent: reads and writes comments in a spec and can turn a PRD into designs, prototypes and code. 5. Agent handoff: paste the spec link into a coding agent; Figma MCP exposes the spec and linked context.

Key Insights

  • The launch rests on the view that deciding what to build, not generating code, is the bottleneck; Figma's design chief cited 17x more code but only 1.3x more releases.
  • Placing the spec on the canvas keeps requirements beside the frames and code layers they describe.
  • Agents take part through ordinary comment threads, which keeps their suggestions reviewable by the team.
  • The Figma MCP server lets agents such as Codex read a spec and its linked Markdown, frames, diagrams, code layers and Code Connect snippets.
  • Code layers entering open beta with an Extract designs action closes the loop from code back to editable design layers.
  • The MCP update adds 20 new integrations and lowers errors and token usage, per Figma.
  • Payload, the headless CMS Figma acquired, is being connected to Figma Make, per TechCrunch.

Was this review helpful?

Share

Twitter/X
Advertisement