> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gooseworks.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart: make three ads from Claude

> From nothing installed to three finished ads, in about five minutes. The literal prompts to type and what comes back.

This is the whole thing end to end: connect Goose to Claude, then ask it to make three ads for your brand. **About five minutes**, most of it waiting on image generation.

You need a **GooseWorks account** ([sign up free](https://make.gooseworks.ai)) and either **Claude Code** or the **Claude app**. Nothing else.

<Steps>
  <Step title="Connect Goose (60 seconds)">
    <Tabs>
      <Tab title="Claude Code (terminal)">
        Paste this into your coding agent — Claude Code, Cursor, or Codex — and let it run the install:

        ```text theme={null}
        Install Gooseworks: run `npx gooseworks@latest install` with the flag for my agent (--claude, --cursor, or --codex). If a browser doesn't open, print the sign-in URL. When it's done, run `npx gooseworks doctor` to confirm the MCP tools are available, then tell me to restart the agent. After restart, my first GooseWorks request should start any unfinished onboarding automatically. When the taste step is reached, open the visual GooseWorks taste-picker page instead of listing the cards in the terminal.
        ```

        The agent runs the install itself, so it can print the sign-in URL if no browser opens, re-check with `doctor` until the MCP tools show up, and continue straight into onboarding. No exact onboarding command is required. **Restart your agent when it's done** — MCP servers are only read at startup.

        A browser opens for Google sign-in. Use the same account as your GooseWorks login. The agent reports the connection verified (`✓ Connected — 131 GooseWorks tools available`), then **quit and reopen Claude Code** — it reads MCP servers only at startup.

        <Accordion title="Manual install (terminal)">
          ```bash theme={null}
          npx gooseworks install --claude
          ```

          Then run `npx gooseworks doctor` to verify, restart Claude Code, and ask GooseWorks to set up your company or begin a real task. Onboarding starts automatically if needed.
        </Accordion>
      </Tab>

      <Tab title="Claude app (claude.ai)">
        Open [claude.ai/new#settings/customize-connectors](https://claude.ai/new#settings/customize-connectors) → **Add custom connector**.

        Name it `Gooseworks` and paste:

        ```
        https://mcp.gooseworks.ai/mcp
        ```

        Click **Add**, then **Connect**, sign in with Google, and click **Allow**. Full walkthrough with screenshots: [Claude](/connections/claude).
      </Tab>
    </Tabs>
  </Step>

  <Step title="Check it's really connected">
    Type this into Claude:

    > **You:** Use GooseWorks to tell me which account I'm signed in as.

    Claude calls `whoami` and answers with something like:

    > **Claude:** You're signed in to GooseWorks as **[you@yourcompany.com](mailto:you@yourcompany.com)**.

    If Claude says it has no GooseWorks tools, it hasn't been restarted — quit it fully and reopen. Still nothing? Run `npx gooseworks doctor`, which tells you exactly which piece is missing.
  </Step>

  <Step title="Give Goose your brand">
    Goose makes ads *for a brand*, so the first thing it needs is your website. It reads the site, works out what you sell and how you talk, and saves that as a reusable brand.

    > **You:** Use GooseWorks to set up a brand for yourcompany.com.

    Claude calls `list_ad_brands` to check what already exists, then `create_ad_brand` with your URL, and finally polls until the research finishes:

    > **Claude:** I've created the brand **Your Company** and researched the site. Research is complete and a brand kit is ready — logo, colours, and a set of messaging angles. Want to see the angles before I make anything?

    <Note>
      This takes 30–60 seconds. It only happens once per brand — every later request reuses it.
    </Note>

    Already have brands in GooseWorks? Skip this. Say *"use my existing Acme brand"* and Claude will find it with `list_ad_brands`.
  </Step>

  <Step title="Ask for three ads">
    Now the actual request:

    > **You:** Make three ad creatives for that brand. Different angles — one problem-first, one product-hero, one social-proof. Show me the credit cost before you spend anything.

    Claude picks starting templates (`surprise_me_templates` for an auto-pick, or `search_ad_templates` if you named a style), then prices the batch **before** generating:

    > **Claude:** I picked three templates that fit your brand. Generating all three will cost **\~120 credits** (you have 13,328). Shall I go ahead?

    > **You:** Yes, go.

    Claude calls `submit_remix_batch` and polls `get_remix_batch` until they finish:

    > **Claude:** All three are done. Here they are:
    >
    > 1. **"Still doing this the hard way?"** — problem-first, 1080×1080
    > 2. **"One tool. Every channel."** — product hero, 1080×1350
    > 3. **"Why teams switch"** — social proof, 1080×1080
    >
    > Each one is linked above and saved to your GooseWorks account.

    <Note>
      **Credits are always priced before they're spent.** Goose calls `estimate_remix_batch`
      first and shows you the number. Ask *"how many credits do I have?"* any time — that's
      `get_ad_credits`.
    </Note>
  </Step>

  <Step title="Change one and re-roll it">
    The point of driving this from a chat is that editing is a sentence, not a UI trip.

    > **You:** The second one's headline is too long. Shorten it to four words and make the background darker.

    Claude calls `regenerate_creative` on just that one and returns the new version. Repeat until you like it.
  </Step>
</Steps>

## What just happened

```
You  ──▶ Claude  ──MCP──▶  GooseWorks
                            ├─ create_ad_brand      read your site, build a brand kit
                            ├─ surprise_me_templates pick starting ads
                            ├─ estimate_remix_batch  price it before spending
                            ├─ submit_remix_batch    generate on GooseWorks' servers
                            └─ regenerate_creative   edit a single one
```

Generation runs on **GooseWorks' infrastructure**, not your machine — Claude is only the interface. Everything it makes is saved to your GooseWorks account and shows up in the [web app](https://make.gooseworks.ai) too.

## If something goes wrong

| What you see                                                 | What it means                                 | Fix                                                                    |
| ------------------------------------------------------------ | --------------------------------------------- | ---------------------------------------------------------------------- |
| Claude says it has no GooseWorks tools                       | The app hasn't reloaded its MCP config        | Fully quit and reopen it                                               |
| `npx gooseworks doctor` says **Not signed in**               | No saved credentials                          | `npx gooseworks login`                                                 |
| `doctor` says **token rejected (HTTP 401)**                  | Credentials are stale or from another account | `npx gooseworks logout && npx gooseworks login`                        |
| `doctor` says **No AI client has the MCP server registered** | Install didn't write the config               | `npx gooseworks install --claude`                                      |
| Sign-in never completes in the terminal                      | The browser callback is blocked               | Retry off your VPN; the CLI also prints a URL you can paste manually   |
| "Brand research failed"                                      | The site blocked the crawler                  | Tell Claude the key facts directly — *"we sell X to Y, our tone is Z"* |

`npx gooseworks doctor` prints the fix next to every failure, so run that first.

## Next

<CardGroup cols={2}>
  <Card title="All the tools" icon="wrench" href="/concepts/goose-growth-tools">
    Every tool the connector exposes, generated from the server itself.
  </Card>

  <Card title="Other AI apps" icon="plug" href="/connections/overview">
    Cursor, Codex, ChatGPT — same account, same tools.
  </Card>
</CardGroup>
