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

# Use cases

> Find example prompts for writing, optimizing, tracking, and reporting with Surfer from your assistant.

export const Prompt = ({text}) => {
  const copyIcon = '<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><rect width="14" height="14" x="8" y="8" rx="2" ry="2"></rect><path d="M4 16c-1.1 0-2-.9-2-2V4c0-1.1.9-2 2-2h10c1.1 0 2 .9 2 2"></path></svg>';
  const checkIcon = '<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M20 6 9 17l-5-5"></path></svg>';
  const copy = event => {
    const button = event.currentTarget;
    if (!navigator.clipboard) return;
    navigator.clipboard.writeText(text).then(() => {
      button.innerHTML = checkIcon;
      button.setAttribute('aria-label', 'Copied');
      clearTimeout(button.resetTimer);
      button.resetTimer = setTimeout(() => {
        button.innerHTML = copyIcon;
        button.setAttribute('aria-label', 'Copy prompt');
      }, 2000);
    }).catch(() => {});
  };
  return <div className="surfer-prompt not-prose">
      <div className="surfer-prompt-meta">
        <svg xmlns="http://www.w3.org/2000/svg" width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round" aria-hidden="true">
          <path d="M19 21v-2a4 4 0 0 0-4-4H9a4 4 0 0 0-4 4v2" />
          <circle cx="12" cy="7" r="4" />
        </svg>
        <span>Prompt</span>
      </div>
      <div className="surfer-prompt-bubble">
        <p className="surfer-prompt-text">{text}</p>
        <button type="button" className="surfer-prompt-copy" aria-label="Copy prompt" onClick={copy}>
          <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round" aria-hidden="true">
            <rect width="14" height="14" x="8" y="8" rx="2" ry="2" />
            <path d="M4 16c-1.1 0-2-.9-2-2V4c0-1.1.9-2 2-2h10c1.1 0 2 .9 2 2" />
          </svg>
        </button>
      </div>
    </div>;
};

Each prompt is written the way you would type it. Under it: what comes back, what it costs, and which capabilities it uses. Use an active workspace for content tasks. If you have not connected Surfer yet, start with the [Quickstart](/mcp/quickstart).

Name the workspace and the target keyword in each prompt. Without both, the assistant has to guess or ask:

<Prompt text="In the Acme workspace, create a Content Editor for 'best CRM for small businesses'." />

Where a prompt below spends a credit, its outcome says so in plain words; [Credits and limits](/mcp/credits-and-limits) lists every action that charges.

<Note>
  For the six core workflows, ask for a [skill](/mcp/skills) instead of
  prompting each step.
</Note>

Tools: [workspaces and recommendations](/mcp/tools/index#workspaces-and-brand).

## Content creation

Create Content Editors and generate articles from an outline you approve.

<Prompt text="Create a Content Editor for '[keyword]' in the [workspace] workspace, then start an AI Article and pause so I can approve its outline before you write." />

What happens, in order:

1. Surfer creates the Content Editor for one Content Editor credit and analyzes the SERP.
2. Your assistant waits for the editor's analysis to finish, then starts an AI Article with outline review enabled.
3. Starting the article reserves a separate AI Article credit if your plan requires one. [Credits and limits](/mcp/credits-and-limits#tools-that-consume-credits) explains when generation is included in the Content Editor allowance.
4. The article pauses once its outline is drafted. You review and edit that outline.
5. Writing resumes from the version you approve.

For planning without starting an AI Article, ask to read the Content Editor's SERP outline. Reading it costs no additional credit. It is separate from the outline prepared during AI Article generation.

<Prompt text="Create Content Editors for these 10 keywords from our Q3 content plan." />

One Content Editor per keyword, one Content Editor credit each; reusing the same retry key keeps a retried create from making a duplicate. See [retry guidance](/mcp/credits-and-limits#timeouts-and-long-running-tools).

For an article you asked to pause for outline review:

<Prompt text="Show me the AI Article outline for [keyword], swap sections 3 and 4, add an FAQ section, then generate." />

Your assistant:

1. Reads the outline once the article is waiting for your input.
2. Sends the edited version back as Markdown.
3. Checks the article until it is complete.

Tools: [Content Editors](/mcp/tools/index#content-editors), [AI Articles and Auto-Optimize](/mcp/tools/index#writing-and-optimizing).

## Optimization and guidelines

Load existing content into a Content Editor, read what the guidelines say is missing, and apply the changes.

<Prompt text="Here's our current blog post on [topic]. Load it into a Content Editor for its main keyword [keyword] and tell me what's missing versus the SEO and AI Search guidelines." />

Surfer creates a Content Editor with the page imported from its URL, for one Content Editor credit. Pasted text is written into the editor as a replacement of its body instead. Once the editor has finished analyzing, you get:

* How many times each recommended term appears in the draft, where zero means absent.
* The facts the SERP and the AI engines cite for the keyword.

From those, your assistant names the ones your draft lacks.

<Warning>
  Auto-Optimize changes the document with no review step; read it back before
  you publish. [AI Articles and
  Auto-Optimize](/mcp/tools/index#writing-and-optimizing) describes what it
  edits.
</Warning>

<Prompt text="Run Auto-Optimize on the following pages: [url list] and show me the content scores before and after." />

Each page needs a Content Editor that has finished analyzing. A page not yet in Surfer is imported into a new editor first, for one Content Editor credit each. Then, for each page, your assistant:

1. Reads the content score.
2. Runs Auto-Optimize for one Auto-Optimize credit.
3. Waits until the job reports that it optimized the page or found nothing to optimize.
4. Reads the score again.

<Prompt text="Which SEO guideline terms haven't I used yet in this draft? Add the top 10 naturally." />

You get the working set of recommended terms with how many times each one appears in the draft. Adding the terms replaces the whole document body and starts a content score recalculation.

<Prompt text="What's my AI Search score for this draft, and which cited information am I missing?" />

* You get the AI Search score, its status, and the number of facts gathered; your assistant compares the facts against your draft.
* Each fact carries its source URL and where that URL was found: the SERP, Google AI Overviews and AI Mode, Gemini, OpenAI, or Perplexity.

Tools: [Content Editors](/mcp/tools/index#content-editors), [guidelines](/mcp/tools/index#guidelines), [AI Articles and Auto-Optimize](/mcp/tools/index#writing-and-optimizing).

## AI Tracker

These prompts need an AI Tracker project in the workspace; your assistant lists the projects to find it. What every read shares:

* Refresh: reports refresh once a day.
* Window: choose a supported report period. The assistant can tell you which dates have reports.
* Model families: Google AI Mode, Google AI Overviews, OpenAI, Perplexity, or Gemini. The summary covers every family at once; the other reads cover one family or all of them combined.

<Prompt text="How visible is our brand in AI search this month?" />

You get the mention rate, average position, and presence score for each AI model family over a 30 day window. A combined row covers all of them.

<Prompt text="Compare our AI visibility against [competitor], broken down by platform." />

For each model family, you get the brands the AI answers mention, sorted by presence score. You can also compare how many cited webpages mention the competitor and how many mention both brands. These are different measures: one describes the AI answers, the other describes the source pages.

<Prompt text="Which of our pages get cited most in AI answers?" />

You get one row per cited URL, sorted by how often it is referenced, with a flag for whether the crawled webpage mentions your brand. An unavailable flag means Surfer does not have that information, not that the brand is absent. The assistant keeps the rows on your domain.

<Prompt text="Show our AI visibility trend over the last 30 days." />

You get the daily mention rate, average position, and presence score across the 30 day window, oldest first. Days without a report are skipped, so the series can be shorter than the window.

<Prompt text="Find third-party sources that mention competitors without mentioning us, verify the pages, and draft an outreach email." />

1. Your assistant compares source mention counts to find competitors with a gap worth investigating.
2. It gets candidate URLs from the sources report. The counts do not identify which URLs mention each competitor; the source report only flags whether each webpage mentions your tracked brand.
3. It inspects candidate pages to verify that they mention the competitor and omit your brand before proposing an outreach list.
4. It drafts the email. No Surfer tool sends it.

Inspecting the webpages needs your assistant's browsing capability. The Surfer reports alone do not produce a verified list of competitor-only sources.

Tools: [AI Tracker](/mcp/tools/index#ai-tracker).

## Recommendations

Read the recommendations, then open a Content Editor for each page worth optimizing.

<Prompt text="List the recommendations for this workspace. Which pages should I optimize first?" />

You get two kinds of recommendation, the optimize block before the write block, each sorted by score from highest to lowest:

* Pages to optimize, from Content Audit: URL, keyword, current position, and previous position.
* Content ideas to write, from topical maps: main keyword, search volume, and difficulty.

Recommendations from AI Tracker mentions are not included; ask for [AI Tracker visibility data](/mcp/tools/index#ai-tracker) directly instead.

<Prompt text="Open the Content Editors for the top recommendations and summarize what needs to change." />

For each page to optimize that has not been started, Surfer opens the page's own Content Editor. That is the same action as the Optimize button in the app.

* It costs one Content Editor credit, unless that editor was already paid for.
* You get each editor's SEO Guidelines: structure targets, competitors, recommended terms, and topics.

Tools: [workspaces and recommendations](/mcp/tools/index#workspaces-and-brand), [guidelines](/mcp/tools/index#guidelines).

## Workspace management

Create a workspace, then set its brand knowledge, templates, and voices.

<Prompt text="Create a workspace from acme.com, activate it when setup is ready, then review the generated brand knowledge with me." />

<Warning>
  Spell the location exactly as Surfer names it, as a full English country name
  such as "United States". A misspelling is not rejected and breaks
  SERP-dependent work later.
</Warning>

Creating a workspace needs:

* An organization owner or admin.
* A name.
* The site URL.
* A location.

What happens, in order:

1. Setup runs in the background and your assistant checks the workspace until it is ready for activation.
2. Your assistant activates the workspace, making it available for brand reads and content work.
3. You get the generated brand knowledge to review. Brand analysis is skipped on plans without brand knowledge.

<Prompt text="Update the brand knowledge in the [workspace] workspace: they've repositioned from 'CRM' to 'revenue platform'." />

Your assistant reads the current brand knowledge and writes the revised version back as Markdown. Only what you change is written, and the brand's site URL cannot be changed this way.

<Prompt text="Create a content template for product comparison pages with these required sections, and a custom voice based on these three sample articles." />

Surfer stores the sections verbatim as the template's reference text and learns the voice's style from the sample articles. Ask to make either one the workspace default. Your assistant checks the reference text against the tools' current input requirements.

Tools: [workspaces and recommendations](/mcp/tools/index#workspaces-and-brand), [templates and voices](/mcp/tools/index#templates-and-voices).

## Reporting

These prompts only read: Content Editor lists, content scores, recommendations, and AI Tracker reports. None of them consumes a credit. Surfer does not schedule reports; if your client can run a saved prompt on a schedule, these are the prompts to schedule.

<Prompt text="What's in progress this week? List all Content Editors with their state and scores." />

You get every Content Editor created since the date you name, each with:

* Its main keyword.
* Its state: scheduled, executing, completed, or failed.
* Its SEO, AI Search, and total scores.

Name a date range and workspace to keep the report focused. Without a workspace filter, the list spans the organization.

<Prompt text="List all editors with their content scores. Which ones score below 70?" />

The list already carries each editor's scores, so your assistant filters it. For an editor whose score is missing or still recalculating, it reads that score on its own.

<Prompt text="What should we work on next month? Summarize the optimize and write recommendations." />

You get both kinds; ask for more to go deeper.

* Pages to optimize carry their current and previous position, so position drops stand out.
* Ideas to write carry search volume and difficulty.

<Prompt text="How did our AI visibility change since last month, per platform? Where's our mention gap versus competitors?" />

* You get a daily trend per AI model family, one family at a time, over a 90 day window.
* The mention gap reports, per competitor, how many cited webpages mention them and how many mention both brands.

Tools: [Content Editors](/mcp/tools/index#content-editors), [workspaces and recommendations](/mcp/tools/index#workspaces-and-brand), [AI Tracker](/mcp/tools/index#ai-tracker).

## Tips

* Name the workspace when you have more than one, or keep a standing instruction in your client about which workspace to use. Otherwise the assistant lists your workspaces and asks.
* Refer to existing editors by keyword. The Content Editor list carries every editor's main keyword, so "the \[keyword] editor" resolves without an id.
* Choose which outline you want to review. The SERP outline is available after editor analysis. Reviewing an AI Article outline requires starting generation first, so any separate AI Article credit is already reserved before approval.
* Chain with your CMS's own MCP server. Your assistant reads the finished document as Markdown or HTML and hands it to the CMS server to publish in the same conversation. No Surfer tool publishes anywhere.

Tools: [workspaces and recommendations](/mcp/tools/index#workspaces-and-brand), [Content Editors](/mcp/tools/index#content-editors), [AI Articles and Auto-Optimize](/mcp/tools/index#writing-and-optimizing).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.