> ## Documentation Index
> Fetch the complete documentation index at: https://mentionscout.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Prompts
> Create and manage the prompts MentionScout runs against AI answer engines.
Prompts are the questions your buyers ask AI answer engines. MentionScout runs each prompt across the engines you choose, then records whether your brand was mentioned, where it ranked, and which sources were cited. The Prompts page is where you create those questions, manage how often they run, and discover new ones to track.
The page has two tabs. **Tracked** is the list of prompts MentionScout runs for you. **Discover** suggests new prompts worth adding. Everything on this page is scoped to your current website.
## The Tracked tab
The Tracked tab lists every prompt for the current website. Two summary cards above the table break your prompts down by **Types** and **Intents** so you can see the mix at a glance.
Each row shows the following columns.
| Column | What it tells you |
| ---------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Avg. vis.** | The share of completed runs that mentioned your brand, drawn as a ring plus a percentage. It reads `n/a` until the prompt has at least one completed run. |
| **Prompt** | The prompt text, linking to the prompt's detail page. A `Paused` tag appears if the prompt is disabled, and an `Excluded from stats` tag appears if it is left out of visibility scoring. |
| **Type** | How MentionScout classifies the prompt: **Organic**, **Competitor comparison**, or **Brand specific**. Shown as a colored letter; hover to read the full label. |
| **Intent** | The buyer intent behind the prompt: **Commercial**, **Informational**, **Branded**, **Comparison**, **Transactional**, or **Local**. |
| **Frequency** | How often the prompt runs automatically. |
| **Runs** | How many times the prompt has run. |
| **Mentioned by** | The answer engines that have mentioned your brand for this prompt, shown as stacked logos. |
| **Created** | When you added the prompt. |
| **Last run** | When the prompt last ran, or `Never`. |
The menu at the end of each row lets you **Run now**, **Open** the prompt, **Edit** it, toggle **Include in stats** / **Exclude from stats**, or **Delete** it.
Deleting a prompt also deletes every run, mention, and citation attached to it. This cannot be undone.
## Create a prompt
Click **New prompt** to add a question. You need a current website selected first; if you have none, the page prompts you to add one.
Enter the question in the **Prompt** field, phrased the way a real buyer would ask an AI assistant. For example, `What are the best AI search visibility tools for SaaS?`
Choose the **Intent**, **Frequency**, **Country**, and **Language** for the prompt.
* **Intent** records the buyer intent (Informational, Commercial, Branded, Comparison, Transactional, or Local) and powers the filters and summary cards.
* **Frequency** controls how often the prompt runs on its own: **Daily**, **Weekly**, **Monthly**, or **Manual only** (never runs automatically).
* **Country** and **Language** set the locale the prompt runs for, so you can track localized answers.
Leave **Exclude from visibility stats** off for normal prompts. Turn it on for prompts that almost always mention you (such as brand-specific questions) so they do not inflate your scores. See [Exclude brand-specific prompts](#exclude-brand-specific-prompts) below.
Under **Run on these LLMs**, select which engines this prompt runs against (for example ChatGPT, Claude, Gemini, Grok, Perplexity, AI Overviews, and AI Mode). By default, every engine you have connected is selected. An engine without an active key shows a `Needs key` badge and cannot be picked.
Click **Create prompt**. MentionScout opens the prompt's detail page, where you can run it and review results.
A prompt's **Type** (Organic, Competitor comparison, or Brand specific) is classified by MentionScout from the prompt text rather than set on the form. It still drives the Type column and filter.
## Edit, pause, and delete
Open a prompt and click **Edit** (or use **Edit** in the row menu) to change its text, run settings, visibility, or answer engines. The edit form adds two controls the create form does not have:
* An **Enabled / Paused** switch. A paused prompt stays in your list and keeps its history but stops running automatically.
* A **Delete prompt** button, with the same confirmation as the row menu.
Click **Save changes** to apply your edits.
## Run prompts on demand
Scheduled runs follow each prompt's frequency, but you can also run prompts immediately.
* **Run now** (row menu) runs a single prompt against every engine you have a key for.
* **Run all** runs every enabled prompt for the current website. Each prompt runs on every connected engine, which queues jobs and consumes API credits.
* **Run N selected** runs only the prompts you tick (see below), including paused ones.
A prompt only runs if it has at least one answer engine selected and an active key for at least one of them. If neither is true, MentionScout tells you what to fix. While runs are in flight, the page shows a live indicator and refreshes the run counts and visibility until they settle. Results land on the [Runs](/ai-visibility/runs) page.
## Filter and find prompts
Use the toolbar above the table to narrow the list:
* The **Filter by prompt text** box matches on the prompt wording.
* The **Type** and **Intent** faceted filters let you focus on one or more categories.
* **Reset** clears any active filters.
* The view options control lets you show or hide columns.
## Bulk actions
Tick the checkbox on one or more rows to reveal bulk actions in the toolbar. They apply to every selected prompt at once.
Replaces (does not merge) the answer-engine list on the selected prompts. The picker defaults to every engine you have a key for; engines without a key are disabled. Click **Apply** to save.
Sets the run frequency (Daily, Weekly, Monthly, or Manual only) on the selected prompts and reschedules their next automatic run.
Choose **Exclude from stats** or **Include in stats** for the selected prompts. They stay tracked and keep running either way; only the scoring changes.
Runs the selected prompts now on every connected engine. This queues jobs and consumes API credits.
## Exclude brand-specific prompts
Brand-specific prompts almost always mention you, so leaving them in your visibility score makes it look higher than it really is. Excluding them gives a truer read of your non-branded visibility, which is where most buyers first encounter you.
You can exclude prompts in three ways:
* The **Exclude from visibility stats** toggle on the create and edit forms.
* **Exclude from stats** in a row's menu, or the **Visibility** bulk action for several prompts at once.
* The **Exclude branded** shortcut, which appears next to **Run all** whenever you have brand-specific prompts still counted. It excludes them all in one click.
Excluded prompts keep running and keep their history; an `Excluded from stats` tag marks them in the list, and your visibility and share-of-voice numbers simply ignore them.
## Discover new prompts
The **Discover** tab surfaces suggested prompts for the current website so you can grow your coverage without writing every question yourself. Review the suggestions and add the ones worth tracking.
Learn how prompt suggestions are generated and how to add them to your tracked list.
## Go deeper
See each individual run, its answer, mentions, and citations.
Expand a prompt into related sub-queries to mirror how engines branch a question.
Find and add new prompts suggested for your website.
Review every place your brand was named across runs.