PandaCrawl
Back to Docs

AI Visibility Agent Skill

Find out whether ChatGPT, Claude, Gemini, Perplexity, Google's AI Overviews and Copilot name your brand, who they name instead, and what to change.

Run this skill in your agent
>/ai-visibility

The AI Visibility Agent Skill measures how AI assistants answer the questions your buyers ask, then turns the answers into a short list of changes. It's the PandaCrawl workflow for generative engine optimisation (GEO).

Your agent picks 8 to 15 questions your buyers would really ask, tracks them across ChatGPT, Claude, Gemini and Perplexity, checks which searches cite you in Google's AI Overviews, checks that your key pages are in Bing, whose index also feeds Copilot and ChatGPT search, and reads how often Copilot cites you if you've uploaded your Bing AI Performance exports.

You get a report that answers three questions: where AI names you today, who it names instead and why, and the three changes most likely to move that.

What this skill helps your agent do

  • Choose buyer questions from evidence: what AI already answers about your competitors, and your own Search Console searches.
  • Set up tracking across four AI models and run it, after showing you the cost.
  • Measure mention rate, position and share of voice against your competitors.
  • Group the sources AI cites into your pages, competitors' pages and third-party pages, and read what they say.
  • Find the Google AI Overviews that cite your competitors but not you.
  • Check Bing indexing for the pages that matter and get missing ones submitted.
  • Read your Copilot citations: the trend, your most-cited pages, and the searches Copilot runs where competitors' pages win.

When to use it

Use this skill when you want to know whether AI assistants recommend you, when a competitor keeps showing up in AI answers, or before and after changes meant to improve how AI describes you. Run it again after a month to compare.

What you get back

A saved report with the verdict, mention rates by model, the questions you're invisible on, who gets named instead, the sources the answers trust, Google AI Overview citations, Bing indexing and Copilot citations, and three changes in priority order, each with the exact page, URL or message.

How to get the best result

  • Save your business and competitors to project context first, or let the skill ask.
  • Give the brand name the way buyers say it, plus any other names it goes by.
  • Approve a weekly schedule so you see a trend, not one sample. AI answers vary from run to run.
  • Connect Bing Webmaster Tools so the skill can check indexing, and upload your AI Performance exports on the Copilot citations page so it can read how often Copilot cites you.

Costs

Reading results, Bing checks, Copilot citations and setup are free. Suggesting questions costs up to about $0.20, each tracking run costs a few cents per question per model, and the agent always shows the run cost and waits for your approval before running it.

Use it with PandaCrawl MCP

Set up PandaCrawl MCP so your agent can run AI visibility tracking, read Google AI Overview citations, check Bing, and read your Copilot citations. See AI Visibility for how tracking works in the app.

Read the actual skill

The source SKILL.md lives on GitHub: