Fullstory MCP

Catch the friction without leaving your workflow

Connect Fullstory MCP in under two minutes and find where your users get stuck, without leaving your AI harness.

Start buildingSee top use cases
✳Claudefullstory mcp · connected

What are the biggest frustrations across our site in the last 7 days? Rank by user count.

✳
Skillfullstory:general-analysis
Fullstory MCP · 3 tools used
  • build_segment
  • compute_metric
  • get_sessions
Top friction, last 7 days:
  1. Rage clicks on “Apply promo” · 1,240 users
  2. Dead clicks on shipping estimate · 610
  3. Card-number field abandons · 420
#1 spiked sharply on Tuesday. Here’s the session behind it.
Session · checkout01:23 · rage ×4 on “Apply promo”Watch →
Agentic Session Review

See the exact moment

Other tools hand your agent a summary of a session. Fullstory hands it the session itself, as structured data rather than raw video, so your agent spends its context on signal instead of footage.

Fullstory MCP
A map first. Then zoom to the moment.

Your agent reads the whole session as structured data, finds the one moment that matters, and opens the exact frame with the DOM behind it.

01:23exact frame
rage ×4
Promo codeCode invalid
Apply
div.promo-field › button.apply[disabled]
The exact frame the user saw, plus the DOM behind it.
Most AI session tools
Pour the whole session in, and hope

Raw video on fast-forward, a frame a second. The agent spends its tokens on footage and still slides past the moment that mattered.

playing 8× · 1 fps
?
A blurry 8x video. The moment that matters blurs right past.

A model watching the video can tell you the screen looked wrong. Reading that same session as data, your agent names the element, the console error, and the network call behind the moment, and it can scan every session instead of the handful someone had time to queue up.

Agentic Session Review is how your agent works a session: it opens the session and read the component tree behind each frame, so it can name the exact moment and the exact element.

Add it

Add the skill from the public repo. fullstory:session-review

Run it every morning

“Find sessions from the last 24 hours that hit an error and dropped out, then review the worst one and tell me what broke.”

Step 1Connect

Connect in a few clicks

Works in the tools you already work in.

Open Connectors
In Claude: Settings › Connectors.
Add Fullstory
Find Fullstory in the Claude Directory and click Add, or choose Add custom connector and paste the URL on the right.
Sign in
Authorize with Fullstory when the browser opens. Sign-in is required.
✳ Claude Directory
Fullstory ✓
Data · Productivity
Add

Behavioral data with receipts. Build segments, compute metrics, then pull the real session behind any number.

Connector URL
https://api.fullstory.com/mcp/fullstory
Sign-in
Required
Add it from the Claude Directory, or paste the connector URL by hand. Full client setup

Authentication and Permission

  • Every person signs in as themselves, in the browser. No keys to copy or store.
  • Your agent reaches only what your own Fullstory account already reaches.
  • Alt: Headless jobs authenticate with an API key as a Bearer token.
Step 2Prompts

Find sessions. Get answers.

The prompt to try first
"What are the biggest frustrations across our site in the last 7 days? Rank by user count."
Top friction: rage clicks on “Apply promo” (1,240 users), dead clicks on the shipping estimate (610), card-field abandons (420). #1 spiked Tuesday.
> Pull three sessions where the promo rage-clicks happened. What were they trying to do?
All three entered a valid-looking code, hit “Code invalid”, retried, then abandoned checkout.
> Open one and show me the screen at the moment it broke.
At 01:23 the promo field read “Code invalid” and the Apply button stayed disabled.Watch this session in Fullstory
Three follow-ups, the way a real investigation runs.
The prompt library now has 40+ prompts, across Session Intelligence, Issue Investigation, User Behavior, Mobile, and Retention. Browse the library
Step 3Skills

Curated expertise to run over and over

This is a curated library of skills from the Fullstory community that we test and maintain, so the answers hold up.

The Skills Library
General Analysis
The right segments, metrics, and funnels for the question, then the number explained with real sessions.
fullstory:general-analysis
Session Review
Drives Fullstory's session tools the way they were built to be: map first, then zoom.
fullstory:session-review
Comparisons
Structures A vs B so the answer means what you think it means.
fullstory:comparisons

“I don't need to learn the UI. I don't need to understand how to create a segment. I can just prompt.”

Product designer
Step 4Routines

Prompt once. Run it daily.

Three example routines that top Fullstory MCP users run today, with the exact prompt behind each.

Catch failures before customers report them

Error logs catch the errors that throw, not the user who rage-clicked a dead button and left.A scheduled agent watches for those silent drop-offs and sizes who is affected.

Users hitting a silent error no log had flagged1,180
Caught on a routine run, hours before the first support ticket.
24/7watch
runs on a schedule, tightened around peak traffic
Beforethe first ticket
surfaced on a scheduled run, not by a customer writing in
The prompt
sre · silent-failure watch
>Every morning, find friction from the last 24h with
no logged error. Rank by users affected.
Ranked silent failures, each with a session link.
Pairs with
Sentry · GitHub

“The MCP flagged something for me that I don't think we would have picked up in any way.”Director of CX and optimization

Anonymized from a real customer interview
Start buildingBrowse the prompt library
Fullstory MCP is in beta. Features and limits are subject to change.