> For the complete documentation index, see [llms.txt](https://docs.fullsession.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.fullsession.io/mcp-server.md).

# 21. FullSession MCP Server

Connect Claude, Cursor, or any MCP client to the hosted FullSession MCP server to query your data in conversation.

The **FullSession MCP server** lets an AI assistant — Claude Desktop, Cursor, or any MCP-compatible client — query your FullSession data directly in conversation. Instead of clicking through the app or writing API calls, you can ask *"which pages had the most rage clicks last week?"* or *"summarize what this user did before they dropped off,"* and your assistant fetches the answer from your live FullSession account.

MCP (Model Context Protocol) is an open standard for connecting AI assistants to external data sources. FullSession ships a hosted MCP endpoint, so there's nothing to install or run yourself.

<figure><img src="https://3931781085-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FjvxfJkYMGfMfN4qlzl0C%2Fuploads%2FoFyxCNEfkZXF35jqlgfq%2Fimage.png?alt=media&amp;token=baef896a-8a1e-41c9-b691-c12068b57040" alt=""><figcaption></figcaption></figure>

***

### 21.1 What the MCP Server Is

The MCP server is a **query and action layer** over your FullSession account. Once connected, your assistant can:

* Search and summarize **session recordings**
* Analyze **heatmaps**, element interactions, and form performance
* Read **funnel** conversion and drop-off data
* Review **Lift AI** goals and recommendations
* Pull **site-wide health overviews** and group-by analytics
* Read **dashboards** and compute ad-hoc charts — metrics, trends, breakdowns, page rankings, feedback, and page flows
* **Create and manage** segments, funnels, saved heatmaps, dashboards, and charts

It sees the same data you see in the app, scoped to the same sites and permissions as your user account.

#### How it differs from the REST API

Both let you get data out of FullSession, but they suit different jobs:

|              | **REST API** ([Chapter 20](https://docs.fullsession.io/api-developer-tools.md)) | **MCP server** (this chapter)                                              |
| ------------ | ------------------------------------------------------------------------------- | -------------------------------------------------------------------------- |
| **Consumer** | Your own code                                                                   | An AI assistant                                                            |
| **Auth**     | API token (`Bearer fus_…`)                                                      | Your FullSession login, via OAuth                                          |
| **Surface**  | One endpoint (list sessions)                                                    | 48 tools across sessions, heatmaps, funnels, Lift AI, dashboards, and more |
| **Best for** | Pipelines, syncs, scheduled jobs                                                | Ad-hoc investigation and analysis in conversation                          |

> If you want to *build* something on your data, use the REST API. If you want to *ask questions* about your data, use MCP — it exposes far more of the product.

***

### 21.2 Connecting Your AI Client

Connecting takes one URL and a browser login. There's no API key to copy or config file to hand-edit.

<figure><img src="https://3931781085-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FjvxfJkYMGfMfN4qlzl0C%2Fuploads%2FIrCB1zPLLGsZLzENvglh%2Fimage.png?alt=media&amp;token=feef41c5-3590-4b70-a148-cc54c652aaa2" alt=""><figcaption></figcaption></figure>

#### Steps

1. In your MCP client (Claude Desktop, Cursor, etc.), open its **connectors / MCP servers** settings.
2. Add the FullSession MCP endpoint:

{% code overflow="wrap" %}

```
https://app.fullsession.io/mcp
```

{% endcode %}

3. The client will prompt you to authorize. A browser window opens to the **standard FullSession login**.
4. Sign in with your normal FullSession credentials — including **SSO**, if your organization uses it ([Chapter 18](https://docs.fullsession.io/sso.md)).
5. Approve the connection. You're returned to your client, now connected.

#### What happens behind the scenes

FullSession implements **OAuth 2.1 with Dynamic Client Registration**, so your client discovers and registers itself automatically — you never paste a client ID or secret. Login is delegated to FullSession's existing identity provider, which is why SSO works out of the box.

Once authorized, your client holds a **short-lived access token** (refreshed automatically) tied to your user account.

> **Your login, your access.** The connection authenticates *as you*. It can reach exactly the sites your account can reach — no more. If you don't have access to a site in the app, your assistant can't query it either.

#### Verifying the connection

Ask your assistant to run **`whoami`**, or simply ask *"which FullSession sites can you see?"* It should return your email and the sites available to you — a quick confirmation that auth worked and that you're pointed at the right account.

***

### 21.3 Tool Reference

The server exposes **48 tools** grouped by area: **30 read tools** that query your data and **18 write tools** that create, change, or delete things in your account. You don't call these directly — your assistant picks the right one from your question — but knowing what exists tells you what you can ask for.

Write tools are marked in the tables below as **write** (adds something new or makes a reversible change) or **destructive** (deletes, or overwrites a definition with no way to recover the old one). See [section 21.5](#read-and-write-tools) for how they're safeguarded.

#### Identity & sites

| Tool                    | What it does                                                                                                                                                       |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **`whoami`**            | Returns the authenticated account and its accessible sites — useful for debugging auth                                                                             |
| **`list_sites`**        | Lists sites you can access, mapping each **URL → site ID**                                                                                                         |
| **`get_site_overview`** | A one-call health summary: sessions, unique visitors, average duration and page count, rage/dead/error click totals, and top landing pages, countries, and devices |

> **Multi-site tip** — if your account has several sites, name the site by its **URL** in your question. Your assistant will resolve it to the right site ID via `list_sites`.

#### Segments

| Tool                                 | What it does                                                                                                         |
| ------------------------------------ | -------------------------------------------------------------------------------------------------------------------- |
| **`get_segments`**                   | Lists your saved **segments** ([Chapter 11](https://docs.fullsession.io/segments.md)) and the fields each filters on |
| **`generate_segment_from_prompt`**   | Turns a plain-language audience description into a segment filter definition — a preview, not saved                  |
| **`create_segment`** *(write)*       | Saves a new named segment on a site                                                                                  |
| **`update_segment`** *(destructive)* | Replaces a segment's title, description, and filters                                                                 |
| **`delete_segment`** *(destructive)* | Permanently deletes a segment                                                                                        |

**`get_segments`** can filter by a name substring and returns up to **100 segments per page** with a cursor. Each segment is flagged if it can be used as the audience for `get_page_heatmap`, so heatmap numbers match what the dashboard shows for that segment.

**`generate_segment_from_prompt`** takes a description like *"mobile users from paid search who rage-clicked on checkout"* and returns a filter definition built with the same generator the app uses — it knows your site's custom events and attributes. Nothing is saved; you can review it, apply it in the segment builder, or pass it on to `create_segment`.

**`create_segment`** expects filters produced by `generate_segment_from_prompt` rather than hand-written ones. Before saving, it runs the real filter → query conversion and **rejects filters that don't resolve to anything**, so you don't end up with a segment that silently matches zero sessions.

**`update_segment`** is a **full replace**, not a partial edit — title, description, and filters are all overwritten. To rename a segment, your assistant re-sends the current filters from `get_segments`. **`delete_segment`** removes the segment permanently; any alert, dashboard, or Lift AI goal built on it loses that reference.

#### Sessions

| Tool                     | What it does                                                                         |
| ------------------------ | ------------------------------------------------------------------------------------ |
| **`search_sessions`**    | Finds sessions matching filters, with frustration-signal counts and **replay links** |
| **`get_session`**        | Fetches one session's metadata, signal digest, and a chronological **timeline**      |
| **`aggregate_sessions`** | Groups sessions by a dimension and returns a metric per group                        |
| **`get_user_sessions`**  | A cross-session profile for one identified user, plus their session history          |

**`search_sessions`** supports a rich filter set — browser, OS, device (Desktop/Mobile/Tablet), country, city, landing page, visited URL, referrer, user ID, email, session duration, page count — plus the frustration flags **has rage clicks / dead clicks / error clicks / JS errors / network errors / abandoned form**. Results can be sorted by **start time, duration, rage clicks, or errors**, and return up to **50 sessions per page** with a cursor for the next page.

**`get_session`** returns a summary and timeline by default (the cheapest answer to *"what happened here?"*), and can optionally include raw **pages, errors, clicks, network, forms, inactivity, custom events,** and **feedback** sections when you need detail.

**`aggregate_sessions`** groups by **country, city, device, browser, OS, landing page, exit page, referrer, or user**, and reports **session count, average duration, total rage clicks, total errors,** or **sessions with errors** — optionally **compared against the previous period**, so you can ask what changed.

<figure><img src="https://3931781085-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FjvxfJkYMGfMfN4qlzl0C%2Fuploads%2F6ReMmX9BnQw8ULYKQHG7%2Fimage.png?alt=media&amp;token=dc9f6fa5-89b4-4b8a-b307-de45936beb6f" alt=""><figcaption></figcaption></figure>

#### Heatmaps & clicks

| Tool                                 | What it does                                                                                                           |
| ------------------------------------ | ---------------------------------------------------------------------------------------------------------------------- |
| **`get_page_heatmap`**               | Aggregated interaction data for a page in **clicks**, **scroll**, or **movement** mode                                 |
| **`get_element_interactions`**       | Deep-dive one element (by CSS selector or visible text) — stats plus the sessions where it was rage/dead/error clicked |
| **`get_form_analytics`**             | Per-form impressions, completion, and abandonment rates                                                                |
| **`compare_pages_by_frustration`**   | Ranks pages site-wide by rage/dead/error clicks, with affected-session counts. Figures match each page's heatmap tiles |
| **`get_heatmaps`**                   | Lists your **saved heatmap views** — the page/device/segment/time-range combinations bookmarked in the app             |
| **`create_heatmap`** *(write)*       | Saves a page/device/segment/time-range combination as a named heatmap view                                             |
| **`update_heatmap`** *(write)*       | Changes one or more fields of a saved heatmap view                                                                     |
| **`delete_heatmap`** *(destructive)* | Permanently deletes a saved heatmap view                                                                               |

These mirror the heatmap features in [Chapter 8](https://docs.fullsession.io/heatmaps.md), but in queryable form — `compare_pages_by_frustration` in particular is a fast way to ask *"where is my site hurting most?"* without opening a heatmap per page. Its page totals and `get_page_heatmap`'s page totals use the same counters as the heatmap's **Total / Rage / Dead / Error** tiles, so the numbers agree with what you see when you open that page's heatmap. Rage clicks are counted **per rage-click burst**, not per individual click.

**`get_page_heatmap`** can be narrowed to a **device** and a saved **segment**; in **clicks** mode it ranks elements by frustration, in **scroll** mode it returns depth milestones and fold data, and in **movement** mode it returns mouse density zones and hotspots. **`get_element_interactions`** reports what share of the page's sessions interacted with the element and lists the sessions where it was rage/dead/error clicked, paginated with replay links. **`get_form_analytics`** works at the form level — field-by-field breakdowns aren't recorded.

Saved heatmap views are bookmarks, not data: **`get_heatmaps`** tells you what's saved, and your assistant computes the numbers with `get_page_heatmap` using the same page, device, segment, and window. **`create_heatmap`** takes a title, a page path (with `*` wildcards), a device (**desktop, tablet, or mobile**), a base segment (or **Everyone**), a time range — either a relative window like *"last 30 days"* or fixed start/end dates — and optionally a funnel to scope it to sessions that reached that funnel. **`update_heatmap`** is a **partial** edit: fields you don't mention stay as they are. **`delete_heatmap`** removes only the saved view; the underlying session data is untouched.

#### Funnels

| Tool                                    | What it does                                                                                   |
| --------------------------------------- | ---------------------------------------------------------------------------------------------- |
| **`list_funnels`**                      | Funnels on a site, with step names, base segment, and processing status                        |
| **`get_funnel_analysis`**               | Per-step users, drop-off, and median effort, plus overall conversion                           |
| **`get_funnel_sessions`**               | Sessions that **dropped at** or **converted through** a step, with replay links                |
| **`generate_funnel_steps_from_prompt`** | Turns a plain-language, numbered list of steps into a funnel definition — a preview, not saved |
| **`create_funnel`** *(write)*           | Saves a new funnel and starts its initial processing                                           |
| **`update_funnel`** *(destructive)*     | Replaces a funnel's name, base segment, and steps                                              |
| **`delete_funnel`** *(destructive)*     | Permanently deletes a funnel and its computed results                                          |

`get_funnel_analysis` can **break down** results by country, device, browser, or OS, or **compare saved segments** against the funnel's audience — the same analysis described in [Chapter 12](https://docs.fullsession.io/funnels.md). `get_funnel_sessions` can also be narrowed to a saved segment; it works on single-session funnels.

**`generate_funnel_steps_from_prompt`** works best with **numbered steps**, each naming one thing that happened — *"step 1: visited /pricing, step 2: clicked Start trial, step 3: visited /thank-you"*. Steps can be page visits (exact or *"any page under /product/"*), clicks (including rage/dead/error clicks), form interactions, errors, or your site's **custom events** by their exact name, and can carry inline conditions such as *"when the device is mobile"*. By default all steps must happen in one session; it can also build **multi-session** funnels that follow a user across visits. It returns the step names for you to confirm, plus the definition to save.

**`create_funnel`** takes a name, a base segment (or **Everyone**), and the generated steps, plus optional default breakdown and up to **4 comparison segments**. Because it can only shape-check the steps, confirm the step order before saving. **`update_funnel`** is a **full replace** of name, segment, and steps. **`delete_funnel`** removes the funnel along with its precomputed results.

#### Lift AI

| Tool                              | What it does                                                                                  |
| --------------------------------- | --------------------------------------------------------------------------------------------- |
| **`list_ai_goals`**               | Your Lift AI goals, each with its latest conversion prediction                                |
| **`get_ai_recommendations`**      | A goal's recommendations — what to fix, rationale, page, step range, and predicted lift       |
| **`get_recommendation_sessions`** | The evidence sessions behind a recommendation, split by affected/unaffected and converted/not |

This makes Lift AI's findings ([Chapter 14](https://docs.fullsession.io/lift-ai.md)) conversational — you can ask your assistant to summarize the top recommendations and explain the evidence.

**`list_ai_goals`** can filter by status, funnel, prediction confidence, or KPI, and flags goals with a run still in progress. **`get_ai_recommendations`** can be narrowed to a step range or a minimum predicted lift, and includes the latest run summary alongside each recommendation's 2×2 evidence counts. **`get_recommendation_sessions`** returns the sessions behind those counts, with replay links.

#### Dashboards & charts

| Tool                                       | What it does                                                                                 |
| ------------------------------------------ | -------------------------------------------------------------------------------------------- |
| **`list_dashboards`**                      | Dashboards you can see on a site, with their charts and grid layouts                         |
| **`get_dashboard`**                        | One dashboard's full definition — each chart's type, what it measures, and its position      |
| **`get_dashboard_data`**                   | Runs every chart on a dashboard for a time window and returns the numbers the app would show |
| **`list_cards`**                           | The site's **chart library** — every saved chart, whether or not it's on a dashboard         |
| **`get_metric`**                           | Computes a single number (total or average of a measure) — like a **metric** card            |
| **`get_trend`**                            | Computes a measure over time — like a **timeseries** card                                    |
| **`get_breakdown`**                        | Computes a measure grouped by a dimension — like a **breakdown** card                        |
| **`get_page_ranking`**                     | Ranks pages by views or load time — like a **pages** card                                    |
| **`get_feedback_snapshot`**                | Summarizes feedback sentiment and scores per widget — like a **feedback** card               |
| **`get_page_flow`**                        | Maps how users move between pages — like a **sankey** card                                   |
| **`create_dashboard`** *(write)*           | Creates a dashboard, optionally filled with new and/or existing charts in one step           |
| **`update_dashboard`** *(write)*           | Changes a dashboard's name, description, or privacy                                          |
| **`delete_dashboard`** *(destructive)*     | Deletes a dashboard — its charts stay in the library                                         |
| **`create_card`** *(write)*                | Saves a new chart to the library, optionally placing it on a dashboard                       |
| **`update_card`** *(write)*                | Changes a saved chart's title, description, type, or definition                              |
| **`delete_card`** *(destructive)*          | Deletes a chart from the library and from every dashboard showing it                         |
| **`add_card_to_dashboard`** *(write)*      | Places an existing library chart on a dashboard                                              |
| **`remove_card_from_dashboard`** *(write)* | Takes a chart off a dashboard — it stays in the library and on other dashboards              |

These tools cover the dashboards and charts described in [Chapter 9](https://docs.fullsession.io/dashboards.md) and [Charts](https://docs.fullsession.io/dashboards/charts.md). **`get_dashboard_data`** is the quickest way to ask *"what does my weekly KPI dashboard say?"* — it defaults to the last 7 days, or any window you name.

The six **ad-hoc chart tools** (`get_metric`, `get_trend`, `get_breakdown`, `get_page_ranking`, `get_feedback_snapshot`, `get_page_flow`) compute a chart on the fly **without saving anything**, so your assistant can answer a question — or preview a chart before adding it to a dashboard. They share one vocabulary:

* **Measures** — sessions, unique users, identified/anonymous users, events, active time, duration, pages per session, clicks, rage/dead/error clicks, page load time, abandoned forms, feedback counts, and error totals (all errors, or by kind: error logs, uncaught exceptions, unhandled rejections, XHR errors, response errors). Numeric measures can be aggregated as **sum, average, median, or a percentile**.
* **Dimensions** (for grouping) — browser and version, country, region, city, device, OS, screen size, page, landing page, exit page, referrer, user ID, clicked element (selector or text), and error details such as the error message or failing request's URL, method, and status.
* **Series** — optionally scope a chart to a subset of sessions (for example *device is mobile*, *country is Germany*, *landing page starts with /blog*), or define up to **5 named series** to compare side by side.

**`get_trend`** can add a **previous-period comparison** or split into one line per dimension value (top 5 by default, up to 50); it picks hourly/daily/weekly/monthly buckets automatically from the window length. **`get_breakdown`** returns up to **50 buckets**. **`get_page_ranking`** ranks any visited, landing, or exit pages by **most viewed** or **slowest**. **`get_feedback_snapshot`** shows current sentiment (**overview**) or sentiment over time (**trends**), for all feedback widgets or one ([Chapter 13](https://docs.fullsession.io/feedback-surveys.md)). **`get_page_flow`** shows **2–10 journey steps** with **2–10 pages per step**.

**`create_dashboard`** can create a dashboard and place up to **15 charts** in one call — new charts, existing library charts, or both — and can make the dashboard **private** to you. Charts are laid out automatically on the dashboard's 3-column grid unless a position is given. Chart titles don't carry a time window, since the window is chosen when the dashboard is viewed.

A chart can sit on several dashboards at once, so edits and deletes behave differently:

* **`update_card`** changes the chart **everywhere it appears**. Only the fields you mention change, but a new chart definition replaces the old one entirely.
* **`remove_card_from_dashboard`** only takes the chart off one dashboard — reversible with `add_card_to_dashboard`.
* **`delete_card`** removes the chart from the library **and every dashboard**.
* **`delete_dashboard`** removes the dashboard but keeps its charts in the library.

***

### 21.4 What You Can Ask

The value of MCP is that you describe the question in plain language and let the assistant choose the tools. Some patterns that work well:

<figure><img src="https://3931781085-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FjvxfJkYMGfMfN4qlzl0C%2Fuploads%2FrjhOqZk1KEZ3UwtB8DIR%2Fimage.png?alt=media&amp;token=d49cc3f1-910b-4dd8-b51a-6c6a818fb102" alt=""><figcaption></figcaption></figure>

#### Triage and investigation

* *"Which pages had the most rage clicks in the last 7 days?"*
* *"Find sessions from mobile users in Germany that hit JavaScript errors yesterday."*
* *"Walk me through what happened in this session"* (paste a session link or ID)
* *"What did user <jane@example.com> do across her last few visits?"*

#### Analysis and trends

* *"Give me a health overview of my site for the past week."*
* *"Group last month's sessions by device and show how error rates changed vs. the previous month."*
* *"Rank my pages by dead clicks and tell me which ones affect the most sessions."*

#### Conversion work

* *"Show me the checkout funnel's drop-off by step, broken down by device."*
* *"Pull sessions where people dropped between step 2 and step 3, and summarize what they had in common."*
* *"Summarize my Lift AI recommendations and which have the highest predicted lift."*

#### Reporting

* *"What does my Weekly KPIs dashboard say for the last 30 days?"*
* *"Chart daily sessions by device over the last month, compared with the previous period."*
* *"Which landing pages are the slowest to load?"*
* *"Show me how visitors move through the site after landing on the home page."*

#### Building and managing

* *"Create a segment for mobile users from paid search who rage-clicked on checkout."*
* *"Set up a funnel: step 1 visited /pricing, step 2 clicked Start trial, step 3 visited /welcome."*
* *"Save a mobile heatmap of /pricing for the last 30 days."*
* *"Build me a dashboard with total sessions, errors by browser, and the top exit pages."*

Your assistant should show you what it's about to create or change and ask before saving — see [section 21.5](#read-and-write-tools).

#### Chaining is where it shines

The real advantage is multi-step work: an assistant can rank pages by frustration, pick the worst one, pull the element-level breakdown, then fetch and summarize a few affected sessions — a sequence that would take a dozen clicks in the app.

> **Tip** — results include **replay links** back into the FullSession app. When your assistant surfaces an interesting session, open the link to watch it yourself ([Chapter 6](https://docs.fullsession.io/the-session-player.md)). Treat the assistant's summary as a fast triage layer, and the recording as the source of truth.

***

### 21.5 Limits, Safeguards & Privacy

#### Read and write tools

The MCP server includes both **read tools** (querying sessions, heatmaps, funnels, Lift AI, and dashboard data) and **write tools** (actions that create or modify data in your account). Every tool respects the same role-based permissions as the FullSession app — your assistant can only read what your role can view and take actions your role is permitted to perform. Write and destructive tools also need the **Let Daleel make changes** permission (`daleel:actions:use`), see *Access scope* below.

| Kind            | Tools                                                                                                                                                                                              | What to expect                                                                                         |
| --------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ |
| **Read**        | All `whoami`, `list_*`, `get_*`, `search_*`, `aggregate_*`, and `compare_*` tools, plus the two `generate_*` tools                                                                                 | Never change anything. The `generate_*` tools draft a definition but save nothing.                     |
| **Write**       | `create_segment`, `create_funnel`, `create_heatmap`, `update_heatmap`, `create_dashboard`, `update_dashboard`, `create_card`, `update_card`, `add_card_to_dashboard`, `remove_card_from_dashboard` | Add something new, or make a change that only touches the fields given or can be undone.               |
| **Destructive** | `delete_segment`, `delete_funnel`, `delete_heatmap`, `delete_dashboard`, `delete_card`, `update_segment`, `update_funnel`                                                                          | Delete data, or **fully replace** a segment or funnel definition — the old version can't be recovered. |

Every write tool's description tells the assistant that it changes data and to **confirm with you first**; destructive tools require that you've explicitly confirmed the specific item by name. Some clients add their own approval prompt before any write. Either way, read what your assistant proposes before you say yes.

#### Time windows

| Behavior           | Detail                                                                                                                                       |
| ------------------ | -------------------------------------------------------------------------------------------------------------------------------------------- |
| **Maximum window** | **90 days**. A longer request is automatically clamped to the most recent 90 days, and the response says so.                                 |
| **Default window** | Most tools default to a recent window (commonly the last 7 days; session search defaults to the last 24 hours) when you don't specify dates. |

If you want a specific period, say so in your question — *"in October"* or *"the last 30 days."*

#### Result caps and pagination

Tools return bounded pages rather than everything at once: **up to 50 sessions** per search page, **up to 50 groups** per aggregation (the remainder folded into an "other" bucket), **up to 100 segments** per page, **up to 50 buckets** per ad-hoc breakdown, and capped items per section when fetching a single session. Each response carries a cursor, so your assistant can page for more when it needs to.

These caps keep responses fast and readable — but be aware that a question like *"list every session this month"* will be answered in pages, not in one dump.

#### Access scope

Access is bound to **your user account**: your customer account and the sites you're permitted to see. You don't pick tools per connection. Instead, **each tool call is checked against your role, per site**, using the same permission the app checks for that feature:

* A read tool runs only on the sites where your role grants the matching view permission. For example, funnel tools need **view funnels** (or **view conversion analysis**), and session tools need **view sessions** (or **view heatmaps**). Sites where your role lacks it are treated as if you had no access to them.
* A write or destructive tool needs the matching create/edit/delete permission **and** **Let Daleel make changes** (`daleel:actions:use`). That permission keeps AI clients read-only for a role that is allowed to edit by hand. Admins have it by default; basic Users don't ([Chapter 24, section 24.7](https://docs.fullsession.io/daleel.md)).
* If your role grants a tool on none of your sites, the call is refused with a *Permission denied* message naming the permission an admin would need to grant.

The same checks apply to [Daleel](https://docs.fullsession.io/daleel.md), FullSession's in-app assistant, which uses these tools.

#### Managing connections

> **There's no in-app screen yet for viewing or revoking connected MCP clients.** If you need a connection revoked — for example, a device you no longer use — contact FullSession support. Disconnecting the server in your AI client also stops it from being used from that machine.

Note also that the MCP endpoint is **not** covered by the REST API's rate limit ([Chapter 20, section 20.4](https://docs.fullsession.io/api-developer-tools.md)) — the two are separate paths.

#### Privacy considerations

Because MCP hands session data to an AI assistant, apply the same care you'd apply to any tool that reads production data:

* **Your masking rules still apply.** Data excluded or masked at recording time ([Chapter 22 — Privacy, Security & Compliance](https://docs.fullsession.io/privacy-security-compliance.md)) was never captured, so it can't be surfaced here either. Masking is your first line of defense.
* **Session data can contain personal information** — emails, names, page content, custom attributes. Be deliberate about which assistants and workspaces you connect, and follow your organization's policy on sending customer data to AI tools.
* **Connect as the right user.** Because access follows your account, connecting from a highly privileged account grants the assistant broad reach. If your organization uses custom roles ([Chapter 16](https://docs.fullsession.io/team-account-management.md)), consider connecting from an account scoped to the sites that actually need analysis, or one without **Let Daleel make changes** if the assistant only needs to read.

> **The big picture** — the FullSession MCP server turns your account into a queryable and actionable interface for AI assistants. Connect once with your normal FullSession login (SSO included), and your assistant gains 48 tools spanning sessions, segments, heatmaps, funnels, Lift AI, and dashboards — bounded by a **90-day window**, paginated results, and your role's permissions on each site. It's the conversational counterpart to the REST API: ask questions in plain language, get answers with replay links back into the app.

***

> **Next up:** [Chapter 22 — Privacy, Security & Compliance](https://docs.fullsession.io/privacy-security-compliance.md) covers what FullSession records and how to control it — the masking and exclusion rules that govern everything the API and MCP server can surface.

***

<br>
