Since we announced the Fullstory MCP beta, customers have made millions of tool calls and found use cases we never planned for. Those calls come from thousands of people at organizations large and small, and from every role: PMs, designers, engineers, QA leads, CS teams, and a few hands-in-the-dirt CXOs.
Anyone who connects to Fullstory MCP can get value from it, whatever their role or technical ability. We’ve made it simple to connect and ask questions. But if you stop at simple questions, you’re taking a picture at the Disneyland gate and going home.
Here’s how top users get the most out of Fullstory MCP.
Connect in a few clicks
Fullstory is now a native connector in Claude, Codex, and Cursor. Find Fullstory, click Add, and sign in. It also works with any other MCP client your team already uses.
Are you the Fullstory person at your company? Stop getting asked if you can “just pull one more thing” for somebody else. Send your team this link and get them connected to Fullstory MCP.
Warm up with simple questions
Once you’re connected, try these warmups:
What are the biggest frustrations across our site in the last 7 days?
Show me the top rage-clicked and dead-clicked elements across our product pages.
What got worse this week vs. last week? Anything trending up sharply?
Here ’s what teams are doing with answers like these:
A QA lead asked why customers abandoned a key flow and posted the answer to Slack. A PM picked it up and started building the fix that day.
A UX director had their agent dig into a checkout error nobody could pin down. It found a user who retried “ship now” three times and failed every time, which ruled out a double-click bug and pointed at the backend. “There’s things I’ve been able to find that nobody in QA or engineering has been able to find.”
An engineer uses it before a feature ships, to plan what to measure: “What events should I have?”
Our MCP prompt library has 40+ more prompts to borrow and get you thinking.
Can’t someone else do it?
One of the greatest Simpsons episodes features Homer Simpson as a candidate for sanitation commissioner. He ran under the campaign slogan “Can’t someone else do it?”.
You might have thought this to yourself as you’ve watched sessions one by one while still missing errors that slipped by your gaze.
Agentic Session Review fulfills Homer’s campaign promise. Yes, AI can review sessions for you. The agent maps the whole session, zooms to the exact moment it broke, and names the element, console error, and network call behind it. Instead of spending your afternoon watching sessions, let an agent do it overnight. It all runs through Fullstory MCP, with no extra setup.
This is already one of the most common things agents do in Fullstory. More than a third of MCP calls last month went to session-level evidence.
Try it:
Find sessions from the last 24 hours that hit an error and dropped out, then review the worst one and tell me what broke.
Extend it with loops
One morning, a customer’s daily Fullstory routine flagged something strange: nearly every visitor had disappeared, except those on one specific browser. A teammate had shipped an update that quietly broke capture on every other browser. No error fired and no dashboard turned red.
“There was no way we could have spotted that without the daily routine going on,” they said.
That’s the difference between asking a question and setting up a loop. A loop is an agent that runs on a schedule or a trigger. Write the question once, and it keeps answering.
The teams running loops get the most out of Fullstory. Their agents work at 3am and on Saturdays.
Here are four loops you can copy for your team today:
1. The silent-failure watch (hourly)
An agent looks for flows that broke without logging an error, then pages on-call. One team’s version posts to Slack with a cowboy personality, “because it makes it a little more fun to read.”
Every hour, find sessions with friction but no logged error. Page me on anything new, and queue the rest ranked by users affected.
2. The bug scoper (every new ticket)
Feedback comes in with a session link. The agent finds what broke and counts who else hit it, so the ticket arrives already sized for prioritization. “It’s got hit 950 times,” one product owner said. “That’s a great sizing for me. Put it in the ticket.”
For each new ticket, open the customer’s session, find the error behind the complaint, and count who else hit it in the last 30 days.
3. The account health deck (weekly)
An agent scores every account and builds the deck, so your first call goes to the account that needs it most. “This deck is 100 percent built by Claude,” one COO said. “I didn’t touch this.” For a small team, he says, it’s “almost like two extra CS reps.”
Every Wednesday, score each account in my book on 30-day usage, flag drops, and draft talking points for the account deck.
4. The opportunity engine (weekly)
A designer’s Cursor automation publishes the five highest-impact fixes every week. Her team’s triage meeting turns them into tickets.
Every Monday, compare users who completed our key funnels against those who abandoned them, find where they diverge, and rank the top five fixes.
When an answer is useful twice, turn it into a loop. That’s the whole trick.
Now it’s your turn
The get-started page has everything you need in one place: setup for every client, the prompts, the skills, and the loops. It also has a setup kit, one file you paste into your AI along with “set this up for me.”
Don’t wait at the gate. Connect Fullstory MCP now and start running your first loops. Then join us at #Bash_ on November 10 to see StoryAI Agents and MCP in action, hear from the teams using them, and get a first look at building for visitors who aren’t human.
