MCP Capabilities

MCP Capabilities

Once the connector is live, your assistant can call the Socialpruf tools directly: it reads your teams and accounts, runs the query your question implies, and works with what comes back. Everything below starts as a plain-language request in the chat.

Start with the basics: your numbers

The foundation is straight KPI retrieval. Posts, followers, views, impressions, plays, likes, comments, shares, saves, earned media value, and engagement rate, for any window you name, filtered to the platforms, accounts, or campaigns you care about.

• Totals - “What were our impressions and EMV last month?”
• Breakdowns - the same numbers split per account, per brand, or per campaign, so you can see who actually carried the month.
• Comparisons - one account against another, one campaign against another, or this period against the last, side by side.
• Post-level detail - search your posts by keyword, platform, date, or duration, sort by virality, EMV, or engagement rate, and pull the day-by-day history of any single post to see how it built.
• Audience - age and gender distribution, top countries, and lookalike profiles for an account or a campaign's roster.

Every one of these runs live at the moment you ask, over history that reaches back to account creation.

Visualizations

Both Claude and ChatGPT can chart whatever the connector returns. Ask for a chart and you get one; ask for a different cut and it re-queries and redraws in the same conversation.

• Time series - impressions, plays, followers, or EMV per day across any range, with campaign launches or a competitor spike called out.
• Rankings and splits - EMV by platform, engagement rate by brand, top twenty posts by virality.
• Heatmaps - day-and-hour posting performance, so you can see when your audience actually shows up.
• Audience breakdowns - age and gender distribution as a population chart, or your top countries mapped for a single account or a whole campaign roster.

Say what the chart is for - a board deck, a client email, a quick gut check - and you'll get a version shaped for it instead of a generic plot.

Plugging into the reporting you already run

This is where the connector stops being a novelty. Your AI platform is already holding your other tools; the Socialpruf connector just makes social data one more thing it can reach in the same conversation.

Combine it with your other connectors

Both platforms let you connect multiple sources at once, and a single request can touch all of them:

• Drive, Sheets, and SharePoint - pull last month's numbers and write them into the reporting sheet your team already maintains, in the columns it already has.
• Notion - draft the monthly recap straight into the page where recaps live, following the format of the previous three.
• Slack - post the Monday summary into the channel, with the stats attached and the contextual commentary that usually takes you twenty minutes.
• Paid and web analytics - if those live in a connected sheet or MCP, ask one question that spans organic and paid and get a single answer, not two tabs you eyeball against each other.

Make the format stick

• Claude Projects - create a project for reporting, add the Socialpruf connector, and put your template and definitions in the project instructions. Every chat inside it starts already knowing your format.
• ChatGPT projects and custom GPTs - the same idea: pin the instructions and reference files so the output arrives in your shape rather than the model's.
• Upload last month's report - the fastest version of this. Hand it the previous PDF or doc and say “same thing, this month's numbers.”

Let it run on a schedule

Where your plan supports scheduled tasks, the recurring report becomes something that arrives rather than something you remember to build. Because the connector reads live data, each run reflects the real state during a Monday morning summary or week over week comparison.

Practical patterns

• The client recap - totals and period-over-period change, the top five posts with why they worked, a chart of impressions over the month, and three lines of narrative.
• The weekly standup note - what moved, what stalled, what to do about it, dropped into Slack before the meeting.
• The pitch or renewal deck - campaign results, audience demographics, and EMV pulled in seconds.
• The anomaly check - “anything unusual in the last two weeks?” and let it find the dip you hadn't noticed, then trace it to the posts responsible.

Socialpruf Teams

Everything the MCP does is scoped to the teams you already have defined in Socialpruf. If your login has access to more than one, the assistant will list them and ask which to use before it answers. It will never mix two teams into one number unless asked.

• Agencies and multi-brand groups - keep each client in its own team and every answer stays clean. Ask for a client recap and you get that client's data, with no risk of another account bleeding in.
• Inside a team - the split continues one level down. Accounts, brands, and campaigns are all first-class, so you can compare two brands this quarter, or see how the summer campaign did against the spring one.
• Permissions carry over - the connector runs as you. It sees exactly the teams and accounts your Socialpruf login already has, and nothing else, so handing it to a coordinator doesn't widen their access.
• Name the team up front - open with “using the Northside team, how did last week go?” Make sure the accounts are added to the team on your Manage page, or the MCP won't see them.

Most importantly, you can check its work. Ask where any figure came from and it points you at the exact posts behind it. The numbers aren't just live and complete, they're auditable.