Getting Started

This page covers everything you need to go from zero to your first generated spec.

uSpec supports two rendering paths. This page sets up the Figma-native path (annotations rendered into Figma). The Component Markdown path (single .md spec via the uSpec Extract plugin) has its own walkthrough on the Component Markdown page.

Setup instructions

Install your AI agent, connect Figma, and configure your template library.

Create your first spec

Prompt format, available skills, and tips for better output.

Setup instructions

uSpec connects an AI coding agent to Figma through an MCP (Model Context Protocol) server. Setup depends on which rendering path you plan to use. Figma-native path (this walkthrough). Annotations are rendered back into your Figma file. You need three pieces:

AI agent host: the coding environment that runs uSpec skills (Cursor, Claude Code, or Codex)
Figma MCP: the bridge that connects your agent to Figma files (either Figma Console MCP or native Figma MCP)
Template library: the Figma file containing documentation templates

Component Markdown path. A single .md spec is written to disk. You need two pieces:

AI agent host (same as above)
uSpec Extract Figma plugin: replaces the MCP + template library requirement. Either install from Figma Community, or — if you prefer the latest source — clone the uSpec repo and build it locally from figma-plugin/. The Component Markdown install guide walks through both options.

For the Component Markdown path, follow the agent host setup below, then jump to the Component Markdown install and usage guide.

You only need to complete this setup once. After that, you can generate specs anytime by opening the project in your agent host.

1. Set up your agent host

Cursor
Claude Code
Codex

Cursor is an AI code editor. All uSpec skills run inside Cursor’s chat.

Download Cursor

Go to cursor.com and download the app for your platform.

Open your project in Cursor

uSpec installs into your existing project — you do not need to clone the uSpec repo. Open any folder where you want to use uSpec.

Install uSpec

In the project’s terminal, run:

npx uspec-skills init

Choose Cursor when prompted. The CLI installs all skills into .cursor/skills/ and references into ./references/, then writes uspecs.config.json.

Select the right model

uSpec skills are token-intensive and require a model with high context capacity. In Cursor’s chat, select OpenAI GPT 5.4 High or Opus 4.6 High or above. OpenAI 5.4 High is more economical to use.

Lower-capacity models may truncate instructions mid-skill, producing incomplete or broken specs. OpenAI GPT 5.4 High or Opus 4.6 High is the minimum recommended model.

Verify skills are available by typing @create-voice or @create-component-md in Cursor’s chat. Either should autocomplete to the skill file.

Claude Code is Anthropic’s terminal-based coding agent.

Install Claude Code

Follow the Claude Code installation guide to install the CLI.

Open your project

uSpec installs into your existing project. cd into any folder where you want to use uSpec.

Install uSpec

npx uspec-skills init

Choose Claude Code CLI when prompted. The CLI installs all skills into .claude/skills/ and references into ./references/, then writes uspecs.config.json.

Select the right model

Use OpenAI GPT 5.4 High or Opus 4.6 High or above. uSpec skills are token-intensive and require high context capacity. OpenAI 5.4 High is more economical to use.

Run claude in the project directory. Claude Code should discover the firstrun skill from .claude/skills/firstrun/.

Codex is OpenAI’s coding agent.

Install Codex

Follow the Codex setup guide to install the CLI.

Open your project

uSpec installs into your existing project. cd into any folder where you want to use uSpec.

Install uSpec

npx uspec-skills init

Choose Codex CLI when prompted. The CLI installs all skills into .agents/skills/ and references into ./references/, then writes uspecs.config.json.

Select the right model

Use OpenAI GPT 5.4 High or Opus 4.6 High or above. uSpec skills are token-intensive and may produce incomplete output with lower-capacity models. OpenAI 5.4 High is more economical to use.

Codex should discover the firstrun skill from .agents/skills/firstrun/. Type $firstrun to get started, or use /skills to browse what’s available.

Using more than one agent host? Run npx uspec-skills install --platform <name> for each additional host in the same project. The CLI installs into the host’s directory without touching the others, so you can have .cursor/skills/, .claude/skills/, and .agents/skills/ side by side.

npx uspec-skills install --platform claude-code
npx uspec-skills install --platform codex

2. Set up Figma MCP

uSpec supports two Figma MCP providers. Choose one based on your setup:

MCP providers update their setup instructions frequently. The configuration examples below are a starting point — always check the provider’s documentation for the latest installation steps and options:

Figma Console MCP: docs.figma-console-mcp.southleft.com
Native Figma MCP: github.com/figma/figma-mcp

Figma Console MCP
Figma MCP (Native)

The Figma Console MCP (by Southleft) connects your agent to Figma Desktop via a Desktop Bridge plugin, giving it access to component data, variables, styles, and screenshots.

Get a Figma personal access token

Go to the Figma API settings

Open figma.com/developers/api#access-tokens

Create a token

Click “Get personal access token”, enter a description like Figma Console MCP, and copy the token. It starts with figd_.

Copy the token immediately. You won’t be able to see it again after closing the dialog.

Configure the MCP in your agent

Cursor
Claude Code
Codex

Add the Figma Console MCP to your Cursor MCP configuration:

{
  "mcpServers": {
    "figma-console": {
      "command": "npx",
      "args": ["-y", "figma-console-mcp@latest"],
      "env": {
        "FIGMA_ACCESS_TOKEN": "figd_YOUR_TOKEN_HERE"
      }
    }
  }
}

Add this to your Cursor MCP settings (global or project-level .cursor/mcp.json).

The repo includes a .mcp.json at the project root. Open it and replace the placeholder token:

{
  "mcpServers": {
    "figma-console": {
      "command": "npx",
      "args": ["-y", "figma-console-mcp@latest"],
      "env": {
        "FIGMA_ACCESS_TOKEN": "figd_YOUR_TOKEN_HERE"
      }
    }
  }
}

Claude Code reads .mcp.json from the project root automatically.

The repo includes a .codex/config.toml. Open it and replace the placeholder token:

[mcp_servers.figma-console]
command = "npx"
args = ["-y", "figma-console-mcp@latest"]

[mcp_servers.figma-console.env]
FIGMA_ACCESS_TOKEN = "figd_YOUR_TOKEN_HERE"

Codex reads MCP configuration from .codex/config.toml.

This uses the NPX setup, which gives you all 59+ tools including design creation and variable management. Setup instructions may change between releases — for the latest configuration options, alternative installation methods, and troubleshooting, see the Figma Console MCP documentation.

Connect to Figma Desktop

The Desktop Bridge plugin provides real-time access to your Figma file data. It comes bundled with the Figma Console MCP.

Install the Desktop Bridge plugin

The Desktop Bridge plugin is included in the figma-console-mcp npm package. After running npx figma-console-mcp@latest, find the plugin at node_modules/figma-console-mcp/figma-desktop-bridge/manifest.json. In Figma Desktop, go to Plugins → Development → Import plugin from manifest… and select this file.

Run the plugin

Open the Desktop Bridge plugin in your Figma file. It auto-connects via WebSocket. No special launch flags needed.

Verify the connection

In your agent, type: “Check Figma status”The agent should confirm an active connection.

When the connection is active, your agent can read Figma file data, take screenshots, and access variables and styles.

The official Figma MCP connects your agent directly to Figma with read and write access. No Desktop Bridge plugin required.

Configure the MCP in your agent

Cursor
Claude Code
Codex

Add the native Figma MCP to your Cursor MCP configuration:

{
  "mcpServers": {
    "figma": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/figma-mcp@latest"],
      "env": {
        "FIGMA_API_KEY": "figd_YOUR_TOKEN_HERE"
      }
    }
  }
}

Add this to your Cursor MCP settings (global or project-level .cursor/mcp.json).

Add the native Figma MCP to .mcp.json at the project root:

{
  "mcpServers": {
    "figma": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/figma-mcp@latest"],
      "env": {
        "FIGMA_API_KEY": "figd_YOUR_TOKEN_HERE"
      }
    }
  }
}

Claude Code reads .mcp.json from the project root automatically.

Add the native Figma MCP to .codex/config.toml:

[mcp_servers.figma]
command = "npx"
args = ["-y", "@anthropic-ai/figma-mcp@latest"]

[mcp_servers.figma.env]
FIGMA_API_KEY = "figd_YOUR_TOKEN_HERE"

Codex reads MCP configuration from .codex/config.toml.

The native Figma MCP supports use_figma for running Plugin API JavaScript, get_screenshot for visual capture, and search_design_system for finding components and variables. Setup instructions may change between releases — for the latest configuration options and full tool list, see the Figma MCP documentation.

Verify the connection

Get a Figma personal access token

Open figma.com/developers/api#access-tokens and create a token with read/write scopes.

Test the connection

In your agent, ask it to run a simple Figma command (e.g., “List pages in my Figma file” with a file link). The agent uses use_figma to verify connectivity.

When the native MCP is connected, your agent can read file structure, execute Plugin API scripts, take screenshots, and search your design system.

Test your MCP connection before proceeding. Run a simple command in your agent to confirm Figma access is working. If the MCP isn’t connected, firstrun will fail when it tries to read your template library.

Console MCP
Native Figma MCP

Open the Desktop Bridge plugin in Figma, then ask your agent: “Check Figma status”. You should see an active connection. If it doesn’t connect, see the Figma Console MCP documentation for setup verification.

Ask your agent: “List pages in my file” with a Figma link. The agent should return the page names. If it fails, verify your FIGMA_API_KEY and see the Figma MCP documentation for setup verification.

3. Run firstrun

The firstrun skill configures your Figma template library. Platform selection and skill installation are handled by npx uspec-skills init (Step 1) — firstrun only handles the Figma side.

You only need to run firstrun once per project. It reads your platform and MCP choice from uspecs.config.json and uses them to extract template keys from your Figma library.

Get the template file

uSpec renders documentation using Figma component templates. You need to add the templates to your Figma organization before running firstrun.

Uber designers can skip this — the internal template library is already configured. When firstrun asks for a library link, type “skip”.

Open the Community file

Go to the uSpec Template on Figma Community.

Duplicate to your drafts

Click Open in Figma to add a copy to your drafts.

Publish as a library

Move the file to your team or organization project, then publish it as a library:

Open the file in Figma
Go to the Assets panel
Click the book icon and select Publish library
Confirm the publish

This makes the templates available for uSpec to import.

If you’re working solo or testing, you can skip publishing and use the file directly. The firstrun skill will still work as long as you have the file open.

Run the firstrun skill

Confirm your Figma MCP is connected

If using Figma Console MCP: Open the Desktop Bridge plugin in Figma Desktop for your template library file. If using Figma MCP (Native): Ensure the MCP server is running and your access token is configured.

Run firstrun

Cursor
Claude Code
Codex

In Cursor’s chat, type:

@firstrun

Invoke the skill directly or let Claude auto-discover it:

/firstrun

Invoke the skill with $ or let Codex match it from the description:

$firstrun

Answer the library prompt

The agent asks one question:Library link — Paste the link to your Figma template library (Uber designers can type “skip”).Your Figma MCP and platform were already configured by npx uspec-skills init, so the agent reads them from uspecs.config.json instead of asking again.

Wait for confirmation

The agent extracts component keys from your templates and merges them into uspecs.config.json automatically.

When you see “Setup complete! You are now ready to use uSpec”, your environment and template library are configured.

Create your first spec

Before running a skill, confirm your setup is complete for the path you want to use. Figma-native path:

Your agent host is open in the project where you ran npx uspec-skills init and skills are available
Your Figma MCP is connected (Desktop Bridge plugin for Console MCP, or native MCP server running)
You’ve run firstrun. The template library is configured (uspecs.config.json has your template keys and MCP provider)

Component Markdown path:

Your agent host is open in the project where you ran npx uspec-skills init and skills are available
The uSpec Extract Figma plugin is installed (from Figma Community or via local dev install)
You have a _base.json file produced by running the plugin on a component

Example prompt: Figma-native path

Reference a skill, paste a Figma link, and describe anything the agent can’t infer from the design: states, variants, behavior, or interaction details.

Prompt format showing skill name, Figma component link, and additional context

Cursor
Claude Code
Codex

@create-voice https://www.figma.com/design/abc123/Components?node-id=100:200

This is a primary action button with enabled, hovered, pressed, and disabled states.
The button has a text label and an optional leading icon.

/create-voice https://www.figma.com/design/abc123/Components?node-id=100:200

This is a primary action button with enabled, hovered, pressed, and disabled states.
The button has a text label and an optional leading icon.

You can also skip the / and describe what you need. Claude auto-discovers the skill from its description.

$create-voice https://www.figma.com/design/abc123/Components?node-id=100:200

This is a primary action button with enabled, hovered, pressed, and disabled states.
The button has a text label and an optional leading icon.

Type $ to mention a skill explicitly, or use /skills to browse what’s available. Codex can also match skills implicitly from their description.

Example prompt: Component Markdown path

Reference the skill and point it at the _base.json produced by the plugin. No Figma link is required.

Cursor
Claude Code
Codex

@create-component-md baseJsonPath=~/Downloads/text-field-_base.json

/create-component-md baseJsonPath=~/Downloads/text-field-_base.json

$create-component-md baseJsonPath=~/Downloads/text-field-_base.json

See the Component Markdown guide for the full plugin install, usage, and cost walkthrough.

Available skills

The skills split into two groups based on where the documentation is delivered. Component Markdown (output is a single .md file):

Skill	What it generates
`create-component-md`	Single self-contained `.md` covering API, structure, color tokens, and screen-reader behavior. Ships with the uSpec Extract Figma plugin. Full guide →

Figma-native (output is rendered into the Figma file):

Skill	What it generates
`create-anatomy`	Numbered markers on a component with an attribute table
`create-property`	Variant axis and boolean toggle exhibits with instance previews
`create-voice`	Screen reader specs for VoiceOver, TalkBack, and ARIA
`create-color`	Color token annotations for all elements and states
`create-api`	Property tables with values, defaults, and examples
`create-structure`	Dimensional specs for spacing, padding, and sizing
`create-motion`	Animation timeline and easing spec from After Effects data

create-component-md has its own setup path. Instead of a Figma link, it takes a _base.json produced by the uSpec Extract Figma plugin. See the Component Markdown guide for the plugin install and usage walkthrough.

What happens when you run a skill

The pipeline differs depending on the path.

Figma-native
Component Markdown

Template import

The agent imports the documentation template into your current Figma page.

Content population

Component name, guidelines, properties, and tables are filled in automatically.

Section cloning

Sections are cloned for each variant, state, or property in your component.

Validation

A screenshot is captured and verified. Issues are fixed automatically.

If a documentation frame appears in your Figma file, your setup is working correctly.

Preflight

The orchestrator validates _base.json against the plugin schema and stages it into .uspec-cache/{componentSlug}/.

API dictionary

extract-api runs first and writes api-dictionary.json. The dictionary keeps the three downstream specialists aligned on property names and states.

Parallel fan-out

extract-structure, extract-color, and extract-voice run as three parallel subagents in a single batch.

Reconciliation

The orchestrator compares specialist outputs for typed disagreements and re-dispatches the owning specialist when needed.

Render

The renderer fills a single Markdown template with the reconciled data and writes components/{componentSlug}.md under your working directory.

If a new .md appears under ./components/ in your working directory, your setup is working correctly.

Tips for better output

Be specific about states: Mention all states (selected, disabled, expanded) in your prompt. The agent cannot infer states that aren’t visible in the current Figma frame.
Describe the parts: For complex components, describe the interactive elements (e.g., “This is a tooltip. The bell icon triggers it, and the bubble appears on hover and focus.”)

Best practices

Do (both paths)

Start a new agent session for every prompt. Each skill consumes significant context. A fresh session ensures the model has full capacity and avoids degraded output.
Run on a high-capacity model. Figma-native skills work on OpenAI GPT 5.4 High or Opus 4.6 High. create-component-md uses parallel subagents and needs Opus 4.7 High or better.

Do (Figma-native only)

Confirm the MCP connection before running a skill. For Console MCP, run the Desktop Bridge plugin in Figma. For native MCP, ensure the server is running. Verify the connection in your agent before starting. If the agent can’t reach Figma, the skill will fail.
Run one agent per Figma file at a time. The MCP supports multiple connections, but all agents share one active page and file context. Finish one skill before starting the next.

Don’t (Figma-native only)

Interact with the Figma frame while a skill is running. The agent writes to the frame in multiple steps. Clicking, selecting, or editing mid-generation can break node references and corrupt the output.
Run multiple agents on the same file simultaneously. They override each other’s page navigation and produce corrupted specs. If you need parallel generation, use separate Figma files.

Don’t (both paths)

Reuse a long chat session for multiple skills. Accumulated history reduces available context and degrades output quality. Start fresh each time.

The Component Markdown path does not write to Figma, so the Figma-interaction caveats above do not apply. You can run multiple create-component-md skills in parallel against different components safely.

Setup issues

Skills not loading

Run npx uspec-skills doctor from your project root. It reports any missing skills, missing references, or broken links.

Cursor
Claude Code
Codex

Confirm the project folder is open in Cursor (not a parent directory)
Check that .cursor/skills/ contains the skill folders. If empty, re-run npx uspec-skills install --platform cursor.
Restart Cursor after running npx uspec-skills init so it picks up the new skills.

Confirm you’re running claude from the project root where you ran npx uspec-skills init
Check that .claude/skills/ contains the skill folders. If empty, re-run npx uspec-skills install --platform claude-code.

Confirm you’re in the project directory where you ran npx uspec-skills init
Check that .agents/skills/ contains the skill folders. If empty, re-run npx uspec-skills install --platform codex.

Figma MCP not connecting

Figma Console MCP:

Confirm Figma Desktop is running (not the web version)
Close and reopen the Desktop Bridge plugin in Figma
Make sure only one instance of Figma Desktop is running
Check that your Figma access token is configured correctly in your agent’s MCP settings

Figma MCP (Native):

Confirm the MCP server is running (check your terminal or agent MCP panel)
Verify your FIGMA_API_KEY is set correctly in your MCP configuration
Test with a simple command like “List pages in my file” with a Figma link

For issues with skills or Figma connection, see the Troubleshooting guide.

Getting Started

Out of Figma

In Figma

Resources

Getting Started

Setup instructions

Create your first spec

Setup instructions

1. Set up your agent host

2. Set up Figma MCP

Get a Figma personal access token

Configure the MCP in your agent

Connect to Figma Desktop

Configure the MCP in your agent

Verify the connection

3. Run firstrun

Get the template file

Run the firstrun skill

Create your first spec

Example prompt: Figma-native path

Example prompt: Component Markdown path

Available skills

What happens when you run a skill

Tips for better output

Best practices

Do (both paths)

Do (Figma-native only)

Don’t (Figma-native only)

Don’t (both paths)

Setup issues

Getting Started

Out of Figma

In Figma

Resources

Documentation Index

Setup instructions

Create your first spec

​Setup instructions

​1. Set up your agent host

​2. Set up Figma MCP

​Get a Figma personal access token

​Configure the MCP in your agent

​Connect to Figma Desktop

​Configure the MCP in your agent

​Verify the connection

​3. Run firstrun

​Get the template file

​Run the firstrun skill

​Create your first spec

​Example prompt: Figma-native path

​Example prompt: Component Markdown path

​Available skills

​What happens when you run a skill

​Tips for better output

​Best practices

​Do (both paths)

​Do (Figma-native only)

​Don’t (Figma-native only)

​Don’t (both paths)

​Setup issues

Setup instructions

1. Set up your agent host

2. Set up Figma MCP

Get a Figma personal access token

Configure the MCP in your agent

Connect to Figma Desktop

Configure the MCP in your agent

Verify the connection

3. Run firstrun

Get the template file

Run the firstrun skill

Create your first spec

Example prompt: Figma-native path

Example prompt: Component Markdown path

Available skills

What happens when you run a skill

Tips for better output

Best practices

Do (both paths)

Do (Figma-native only)

Don’t (Figma-native only)

Don’t (both paths)

Setup issues