Skip to main content
These recipes show how to combine Promptwatch MCP tools into useful workflows. Copy the prompts directly into your AI assistant, or use them as starting points for your own.
Each recipe assumes you’ve already connected the MCP server to your AI assistant.

​
Weekly GEO Briefing

Get a structured overview of your brand’s AI visibility — trends, competitor movements, and citation changes — ready to share with your team. What it does: Pulls visibility trends, competitor rankings, and citation data for the past week, then synthesizes them into a shareable summary. Tools used: getVisibilityTimeSeries, getTopCompetitors, getCitations, getSentimentTimeSeries Prompt:
What to expect: Your assistant will call multiple tools, compare this week against last week, and return a structured summary with highlights and lowlights. Takes about 30 seconds.

​
Generate a PDF Report

Create a dashboard or monitor PDF Report and retrieve a Download URL when generation finishes. What it does: Queues a Report with the same filters as the in-app Reports UI, polls until the PDF is ready, then returns a short-lived Download URL. Email is never sent from these tools. Tools used: createReport, getReport (optional: listReports) Prompt:
What to expect: Your assistant calls createReport (returns a reportId immediately), then polls getReport until status is complete and shares the Download URL. Generation is usually quick; if it is still pending, ask again shortly or use listReports to find the row.
For a single monitor, pass type monitor and the monitor id from listMonitors as options.llmMonitorId.

​
Find and Fix Content Gaps

Go from “where are we missing?” to “here’s a draft” in a single conversation. The MCP identifies prompts where your brand isn’t showing up, then generates optimized content to close the gap. What it does: Scans your monitored prompts for low-visibility areas, pulls specific recommendations, and generates a content piece targeting the gap. Tools used: listContentGapPrompts, getContentGapRecommendations, createContent, getContent Prompt:
What to expect: Your assistant will surface prompts where competitors are visible but you’re not, explain what content would help, and kick off AI content generation. You’ll get a full draft article optimized for the gap. Content generation runs async — the assistant will poll until it’s ready.
Follow up with “Now do the same for the second gap” to work through multiple gaps in one session.

​
Publish a Draft and Track It

Close the loop on a generated article: push it to your CMS, take it live, and start measuring whether AI answers begin citing it. What it does: Finds a completed document, stages it in your connected CMS for review, then publishes it live. Publishing automatically adds the resulting URL to your page tracker, so citations to the new page are picked up from that point on. Tools used: listContent, listCmsConnections, pushContentDraft, getContentPublishStatus, publishContentLive, listPageTrackerResponses Prompt:
What to expect: Your assistant lists your CMS connections, pushes the draft, and reports the staged item with a preview link. Once you confirm, ask it to “publish it live” — it promotes the same item and links the live URL to your page tracker.
publishContentLive puts content on your public site and cannot be undone from Promptwatch. Ask your assistant to stage a draft first and confirm with you before going live.
Published outside a CMS? Use “Mark this document as published at <url>” instead — setContentPublication records the URL and starts tracking it without touching a CMS. Check back a week later with “Has anything cited that page yet?”

​
Investigate a Visibility Drop

When your visibility score drops, use this workflow to trace the cause — is it a crawler access issue, a citation shift, or a competitor gain? What it does: Compares visibility before and after the drop, checks if AI crawlers are still reaching your site, and looks for changes in which sources are being cited. Tools used: getVisibilityTimeSeries, getCrawlerTrend, getCitationRankAnalysis, listResponses Prompt:
What to expect: Your assistant will pull time-series data to pinpoint when the drop happened, cross-reference with crawler activity to rule out access issues, and check citation rankings to see if competitors displaced you. You’ll get a diagnosis with specific prompts and models affected.
If the cause is a citation shift, follow up with “Show me what sources replaced us” to see exactly which competitors or domains gained the citations you lost.