ChatGPT Connectors for Figma
You can connect Figma to ChatGPT through a custom connector that reads design data via the Figma API. Here is what ChatGPT can see, what stays outside its reach, and how to record approved design decisions.
Updated 5 min read
You can connect Figma to ChatGPT through a custom connector on the ChatGPT connector platform that reads design data from the Figma API. There is no first-party Figma connector in ChatGPT's standard catalog. Figma does provide an official MCP server that works with AI assistants supporting MCP, but connecting it through ChatGPT's connector platform is a separate step that requires a custom setup. Tools like fenbs keep the approved design decisions and implementation handoffs in a place that survives when the AI session ends.
This page covers the Figma connection options for ChatGPT, what design data is readable, and what to check before you build.
How the Figma API connects to ChatGPT
Figma has a REST API with personal access tokens and OAuth 2.0 for authentication. You can build a custom connector on the ChatGPT connector platform that uses these credentials to read design files. When ChatGPT calls the connector, it sends requests to Figma's API and returns the data as structured JSON: file names, page lists, frame hierarchies, component names, and node properties.
Building a connector requires developer mode in ChatGPT. Which plans include it is OpenAI's to confirm: check OpenAI's help center (opens in a new tab). For the connector registration steps, see chatgpt connectors api.
What ChatGPT can read from Figma
The Figma REST API exposes design file structure as JSON. With a working connector, ChatGPT can read the following.
- File and page structure: the names and IDs of all pages and frames in a file.
- Component names and descriptions: component names, variants, and any description text attached in Figma.
- Text content: the visible text inside frames and components.
- Comment threads: unresolved and resolved comments on a file.
- Version history metadata: who made changes and when, without the actual diff.
What ChatGPT cannot read: rendered visual output. The API returns structural data, not pixels. ChatGPT cannot see what a design looks like visually. If you need ChatGPT to respond to the visual appearance, take a screenshot and upload it separately.
Access limits to check
A few limits affect what a Figma connector can read.
- File access: the authenticated account must have at least View access to a Figma file. Private files in another user's draft workspace are not accessible.
- Token type: a personal access token gives access to files the account owns or has been shared. An OAuth token limits access to the scopes the user approved.
- Organization files: in Figma Organization and Enterprise plans, files can have team-level permissions. Verify that the authenticated account has the right team membership.
- API rate limits: Figma's API has rate limits per token. High-frequency connector calls can be throttled.
Recording approved design decisions
A ChatGPT session that reads Figma data can help review a design, explain component structure, or flag inconsistencies. But the conclusions from that session, what was approved, what needs changing, and which components the developer should use, belong outside the chat. When the next session opens or a developer picks up the handoff, they need the current agreed state, not a transcript.
fenbs keeps those conclusions as Decisions and rules (the approved design choices) and task plans (the implementation steps agreed from the review). The developer's AI session reads those before touching code, so the same approved decisions carry through to the build. For privacy considerations when sharing design data, see chatgpt connectors privacy.
The handoff between design review and implementation is where most design intent is lost. A ChatGPT session that reviews a Figma file can produce a clear list of decisions: this component is the approved navigation pattern, that spacing value is the agreed standard, the mobile breakpoint is set at 375px. But if those decisions stay in the chat transcript, the developer who opens the file the next day has no access to them. They may implement a different navigation pattern, use a different spacing value, or ask the designer a question that was already answered. Writing the approved decisions to fenbs immediately after the review means the developer's AI session starts with the exact choices that the review produced, with no gap between what was agreed and what gets built. It also means the designer does not need to be available for every developer question: the decision record answers the routine ones.
Record approved design decisions once, apply them in every build session
After a Figma review session in ChatGPT, add the approved decisions to fenbs as Decisions and rules: the accepted component, the approved spacing, the rejected approach and why. Connect your developer's AI tool from Integrations and ask it to read the fenbs context before it starts building, so it follows the design decisions without needing the designer in the session.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
How to do this with fenbs
- Create a free account. One person is free with no credit card.
- Add the design project. Write the goal and add the current design tasks.
- After a review session, write the approved design decisions as Decisions and rules: what was accepted, what was rejected, and why.
- Connect ChatGPT (and your developer's AI tool) to fenbs from Integrations.
- Ask each AI session to read the fenbs context before starting. Before stopping, ask it to save progress and record any new design decisions approved in that session.
fenbs does not connect to Figma directly. It keeps the design decisions and handoff notes that bridge design and development. For related connector workflows, see chatgpt connectors and apps and chatgpt connectors not working.