> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.knapsack.cloud/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.knapsack.cloud/_mcp/server.

# Prompting Intelligent Tools

AI features like Intelligent Tools use tokens billed directly by your AI provider. Knapsack doesn't manage these costs.

## Introduction

Once your AI tools have access your Knapsack content it with give live access to your design tokens, components and documentation so that outputs are grounded in the guidelines and standards that define your brand.

> The connection layer and system intelligence — surface what's missing and recommending what's next

***

## I've set up the Knapsack MCP, now what?

With the Knapsack MCP connected, your AI tools can now query your design system in real time — tokens, components, documentation, and all. Instead of hallucinating generic patterns, your AI assistant answers with *your* components, *your* terminology, and *your* standards. Use it to audit gaps, generate design system-aligned output, and get answers grounded in what your team has actually built.

#### Example prompts

> These prompts take advantage of your connected data sources. They work best when you have product, code, or analytics sources synced — the more context you've connected, the more precise the output.

##### Audit existing design system

Let's say you are interested to learn where the gaps within your design system are, you could try a prompt like this:

Tell me all of the gaps within my design system

##### Generate using components

Say you want to generate something using your components, you could try a prompt like this:

Generate a form using our design system

##### Identify and fill documentation gaps

Let's say you are interested in getting help writing missing documentation for a component, you could try a prompt like this:

\[Component] lacks documentation. Please write markdown to document how and why we would use this pattern. Please use all properties available.

Replace `[Component]` with the component you'd like help missing documentation for

##### Plan for the next sprint

You need consultation on how best to manage your teams workload, you can try a prompt like this:

According to what you know about my design system, where should I focus future development work on?

To get even more from this you you could add in resourcing constraints

##### Communicate a design system update

You've shipped updates to your design system and need to help your team understand what changed and why, you can try a prompt like this:

Summarize the most recent changes to our design system in plain language. Explain what changed, what it affects, and what teams need to action.

***

#### Unlock the power of Data Sources

For teams who want more, Intelligent Tools allows you to create a contextual data layer to empower MCP workflows. Contact your sales partner to discuss unlocking data sources.

## I've connected Data Sources, now what?

With Data Sources connected, your AI tools go beyond documentation — they can cross-reference your live product, code repositories, analytics, and more. Now you can catch design drift before it ships, audit adoption at scale, and generate outputs that reflect not just what your system *says*, but how it's actually *used*. Think of it as adding a real-time mirror to your design system intelligence.

#### Example prompts

> These prompts take advantage of your connected data sources. They work best when you have product, code, or analytics sources synced — the more context you've connected, the more precise the output.

##### Keep your sources in sync

You noticed a discrepancy, and are worried it might be happening in other places in the product, you could try a prompt like this:

Compare \[discrepancy] against my connected components and data sources. Flag any inconsistencies, and identify which components are most at risk of rendering incorrectly.

##### Monitor compliance and adoption

Say you want to audit the adherence to your governance policies and processes, you could try a prompt like this:

Audit the design, code, and brand displayed in our live product, and produce a prioritized report for how to address drift.

##### Generate using components

Say you want to generate something using your components, you could try a prompt like this:

Generate a form using our design system

##### Identify and fill documentation gaps

Let's say you are interested in improving the quality of your design system documentation, you could try a prompt like this:

Score my current design system documentation, tell me where the inconsistencies and gaps are, and recommend solutions to solve the inconsistencies and gaps.