Cited By ยท MCP

Use Cases

Example prompts and workflows for the ThriveStack MCP Server. Browse by type if you need ideas on what to ask.

Start by just asking a question — your AI assistant resolves your product and environment automatically from how you connected, so there’s no project to pick first:

“How visible is ourdomain.com in AI search right now?”
Include a time range in your questions. “Last 30 days”, “this month”, or “past week” gets you better results.
For a structured AI-visibility pass instead of one-off questions, invoke the ai_visibility_analyst prompt (see Prompts).

Example workflows: AI Visibility

AI Visibility is the flagship live capability today — these tools cover a full loop from “where are we losing citations” to “here’s a content outline to fix it.”

1. Full visibility snapshot for a domain

“How visible is ourdomain.com in AI search right now?”

get_domain_report returns overall visibility and average position for the domain’s latest completed probe run, broken down by topic and by individual tracked question, with per-AI-model detail underneath each. If the domain hasn’t been probed yet, you get a clear empty result, not an error.

2. Citation gap deep-dive

“Where are we losing AI citations for ourdomain.com, and which competitors are showing up instead?”

get_citation_probe_data returns the raw evidence behind the gaps: every tracked buyer question, your citation count vs. total, and which competitor brands appeared in your place. Use this when you want the underlying evidence rather than get_domain_report’s rollup.

3. See what buyer questions you’re tracked on

“What buyer questions is ourdomain.com being evaluated on?”

get_domain_prompts lists the tracked buyer questions ThriveStack has generated for the domain, grouped by topic — useful context before interpreting a citation gap or requesting content recommendations.

4. Turn a gap into a content plan

“What content should we publish this week to close our biggest AI citation gaps?”

get_owned_content_recommendations returns gap-scored, prioritized recommendations (severity tier, gap score, citations vs. total, competitors winning) for your own site. Pick the top one and call generate_content_outline to get headings, an answer-first summary, FAQ questions, and schema markup suggestions — ready to hand to a writer.

5. Find third-party outreach opportunities

“Where could we get cited on sites like Reddit or G2 instead of just our own blog?”

get_earned_content_recommendations surfaces ideas for earned/third-party media. Each recommendation is flagged citation_backed (tied to a real probe gap) or category_inferred (no direct citation evidence yet) — check which one before treating it as fact.

6. Combine the full picture in one pass

“Give me the full AI-visibility picture for ourdomain.com: how we’re doing, why, and what to publish next.”

The assistant chains get_domain_report (the headline numbers) with get_citation_probe_data (the gap evidence) and get_owned_content_recommendations (what to do about it) — or you can invoke the ai_visibility_analyst prompt (see Prompts) to get this as a single structured pass instead of three separate questions.

7. Change what you track, then run a fresh check

“Suggest five more prompts about pricing for ourdomain.com, track the ones about enterprise plans, and run a new visibility check.”

generate_suggested_prompts writes the suggestions for the topic, track_prompts starts tracking the ones you pick (and returns their new ids), and start_probe_run runs every tracked prompt again. get_probe_run shows progress, and get_domain_report has the results when the run finishes. A run uses your plan’s answer quota, so your AI client may ask you to confirm before it starts. To stop tracking something, archive_prompts moves it to Archived and reactivate_prompts brings it back; see Managing Prompts & Probe Runs.

8. Find the prompts where rivals’ ChatGPT ads beat you

“Which of ourdomain.com’s prompts are losing to rivals’ ChatGPT ads, who is winning them, and what do their ads say?”

get_chatgpt_ads_report gives the market picture first: who advertises and their share. get_chatgpt_ads_prompt_scorecard then lists the prompts by status, losing first, with a count per topic. get_chatgpt_ads_rivals ranks the advertisers that beat you and shows whether any of them bids on prompts that name your brand. get_chatgpt_ads_prompt_detail then shows the ads on one prompt, so you can compare your copy with theirs. Check low_confidence before acting on a prompt with few samples, and use get_chatgpt_ads_runs to see when the data was last refreshed. These tools need the ChatGPT Ads add-on or its trial; get_chatgpt_ads_addon_status says whether the brand has it. For the whole flow, from the report to a drafted ad creative, see the ChatGPT Ads guide.

More ideas

AI visibility
  • Full visibility report for a domain’s latest probe run, with per-model detail
  • Raw citation-gap evidence, including which competitors appeared instead
  • The tracked buyer questions generated for a domain, grouped by topic
  • Owned vs. earned content recommendations, filterable by status, priority, or channel
  • A ready-to-write content outline for a specific recommendation
Brands
  • Which brands/products your account can access
See Tools for everything available today, including the site-health, source and prompt-research tools.

Real-world workflows

Turn an AI-visibility gap into a content plan

“What content should we publish this week to close our biggest AI citation gaps?”

get_owned_content_recommendations returns gap-scored, prioritized recommendations with the evidence behind each; pick the top one and call generate_content_outline to get headings, an answer-first summary, FAQ questions, and schema markup suggestions to hand to a writer.

Prep for a competitive review

“Before our QBR: how does ourdomain.com’s AI visibility compare to last quarter, and which competitors are winning the questions we’re losing?”

get_domain_report gives the headline numbers, get_citation_probe_data names the specific competitor brands appearing in place of yours on each losing question — concrete talking points instead of a single visibility percentage.

Tips

To make the experience smoother or more complete, take a look at the following tips:

  • Include dates: Always specify a time range or month (e.g. “last 7 days”, “October 2026”).
  • Let the assistant discover your schema automatically: For anything referencing your own event or feature names, the assistant figures out your real event taxonomy before querying — you don’t need to know exact event names.
  • Start broad, then drill down: Ask for the overview first, then follow up on whatever stands out.