Promptwatch MCP Server, REST API, Slack, and Looker Studio for AI Visibility Data
Four ways to get AI visibility data out of Promptwatch and into the place where the work happens: a hosted MCP server, a REST API, a Slack agent, and a Looker Studio connector.
A visibility platform that only lives in its own dashboard is a partial platform. The data has to get out, into the place where the work actually happens, which is usually a coding agent, a reporting tool, a chat channel, or a BI dashboard. Promptwatch ships four integration surfaces for that: a hosted MCP server, a REST API v2, a Slack agent, and a Looker Studio connector. This post covers what each one is for, which tools it exposes, and where the data lands.
The tool names below are the real ones. The numbers come from the Promptwatch fact sheet. Where a surface has a plan gate, we say so.
The hosted MCP server
The MCP server is the surface for coding agents and AI assistants. It runs as a hosted streamable HTTP server at https://server.promptwatch.com/mcp, with OAuth or a Bearer key. The ChatGPT plugin and the Claude connector run on the same server, and Cursor connects via .cursor/mcp.json. The point of a hosted server is that you do not run infrastructure. You point a client at the URL, hand it a key, and the agent can call the tools.
The read tools are the ones an agent uses to diagnose. listPrompts and getPrompt read the tracked prompt set and a single prompt in detail. getVisibilityTimeSeries shows how a prompt's visibility moved over time. listQueryFanouts exposes the fan-out structure around a seed query. getCitations, getCitationTopPages, listRedditCitations, and listYoutubeCitations read the citation layer, including the offsite channels. getCompetitorHeatmap shows who shows up alongside you. listContentGapPrompts, getContentGapRecommendations, and getContentGapStats surface the content gap. getCrawlerTrend and getTopCrawlerPages read the AI crawler logs. getVisitorTrend, getVisitorTopPages, and getVisitorTopSources read the AI-referred traffic. getResponseSentimentTimeSeries reads sentiment. listAds and listAdPrompts read the Ads Radar surface. listShoppingItems reads tracked shopping items. getSiteHealth reads structural problems. listActionItems reads the action board.
The write tools are the ones that mutate state. createContent drafts from a gap. listContentSlots reads the review inbox. acceptContentSlot moves a draft forward. publishContentLive pushes the approved draft to the connected CMS. createReport builds a report. createActionItem and updateActionItemStatus manage the action board. createPrompt and createPromptsBulk add tracked prompts. addShoppingTrackedProducts extends shopping tracking.
The scope boundary is the property that makes the MCP server safe to hand to an agent. A read-only key hides the write tools from the client. The agent literally cannot call publishContentLive or createContent, because the client never received them. That is a stronger guarantee than a permission flag, and it is the reason a read-only key belongs in your permanent config and a write key belongs only in the narrow loop you have reviewed.
The REST API v2
The REST API v2 is the surface for scripts, pipelines, and tools that are not MCP clients. It is included from the entry paid plan, the same as the MCP server. The use cases are the ones where you want a programmatic call, not a conversational agent: a nightly script that pulls visibility into a warehouse, a CI check that fails if a tracked prompt drops below a threshold, a custom dashboard that blends Promptwatch data with first-party data the platform does not hold.
The API covers the same data the MCP read tools cover, because it is the same data layer underneath. The difference is the shape. MCP returns tool calls an agent reasons over. The REST API returns JSON a script parses. If you are building a cron job, the REST API is the surface. If you are building an agent loop in Claude or Cursor, the MCP server is the surface. Most teams that run serious programs use both, and the read-only key pattern applies to both: keep a read-only key in the script that reports, and a write key only in the script that acts, gated behind a human.
The Slack agent
The Slack agent is the surface for the team that lives in chat. It connects through an org owner OAuth from Settings, maps channels to projects, and answers an @Promptwatch mention or a direct message. The use case is the one where a teammate asks "did we drop on the pricing prompt this week" and gets the answer in channel, without opening a dashboard.
The Slack agent is the surface that turns visibility data into a shared, conversational artifact. A weekly briefing can be scheduled into a channel, and the briefing skips when nothing changed, which is the part that keeps the channel from becoming noise. The agent reads the same data the MCP server reads, so the answer in Slack is consistent with the answer in the dashboard or in Cursor. The Slack surface is the one that gets visibility data in front of the people who do not log in to the platform every day, which is most of the people who need to act on it.
The Looker Studio connector
The Looker Studio connector is the surface for reporting and BI. It is available on Professional and the agency plans, and it pulls AI visibility data into a Looker Studio report that auto-refreshes, with templates for monitors, citations, prompts, and visibility. The pull window is 90 days, which is the right window for a visibility trend, long enough to see movement and short enough to stay current.
The Looker Studio surface is the one that blends Promptwatch data with the other data a reporting stakeholder needs. The pattern that works is one Looker report with the Promptwatch connector next to native GA4 and native Google Search Console, with a clear rule: do not average AI mentions with GSC clicks. They are different units. AI mentions are a citation count. GSC clicks are a click count. Averaging them produces a number that means nothing. The right report puts them side by side, labeled, so a stakeholder can read the AI surface and the Google surface separately and see how they relate.
The connector is also the surface for client reporting on an agency retainer. A per-project read-only key feeds a Looker report that a client can open without touching the platform, which is the shape a white-label client report takes when the full white-label dashboard is enterprise-gated.
How the four surfaces fit together
The four surfaces are not competing options. They are four doors into the same data, for four different consumers. The MCP server is for the coding agent that reasons over the data and proposes work. The REST API is for the script that moves the data on a schedule. Slack is for the team that asks questions in chat. Looker Studio is for the stakeholder who reads a report.
A mature program uses all four. The agent in Cursor or Claude calls the MCP server to diagnose a visibility drop and propose a fix. A cron job calls the REST API to pull the weekly numbers into a warehouse. The Slack agent posts the briefing into the team channel on Monday. The Looker report feeds the client review on Friday. The data is the same in all four, because it is the same data layer underneath, and the read-only key pattern keeps the write tools out of every surface that does not need them.
The plan gates that matter
The MCP server and the REST API v2 are included from the entry paid plan, Essential at $95 a month. The Looker Studio connector is on Professional at $245 a month and the agency plans. The Slack agent is on the integration list from the paid plans. The free Explore tier, 10 prompts on ChatGPT only, is for evaluation, not for running an integration program. A team that wants to wire the four surfaces into a real workflow is on a paid plan, and the read-only key is the default key in every config that is not the narrow acting loop.
The named feature
The integration surface that closes the loop is the MCP server at https://server.promptwatch.com/mcp, with the read tools that diagnose and the write tools that act, gated behind the read-only key pattern. The REST API v2 is the same data for scripts. Slack is the same data for chat. Looker Studio is the same data for reports. The point of the four surfaces is that the data does not stay in the dashboard. It gets to the agent, the script, the channel, and the stakeholder, and the acting half happens where the work already happens, not in a tab you have to remember to open.