> ## Documentation Index
> Fetch the complete documentation index at: https://chatobserver.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# What can I do with Chatobserver in an AI assistant?

> See every supported Chatobserver MCP workflow, what remains read-only, and which product areas are not yet available.

The Chatobserver connector lets Claude, ChatGPT and other compatible assistants analyse your AI visibility data and, with your permission, make selected workspace changes. You can describe the outcome you want in ordinary language. The assistant chooses the necessary Chatobserver actions and asks for approval when required.

## Available now

| Area                  | The assistant can read and analyse                                                                                                  | With change permission, it can also                                                                                                         |
| --------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------- |
| Workspaces            | List the connected workspaces and the access available in each                                                                      | Nothing; workspace administration stays in Chatobserver                                                                                     |
| Usage                 | Show your plan, active prompt capacity, AI answer allowance and recent usage periods                                                | Nothing; billing and plan changes stay in Chatobserver                                                                                      |
| Prompts               | Search prompts, inspect settings, review the prompt table, compare visibility, inspect popularity and read drafts                   | Create or import prompts, save drafts, edit settings, activate, deactivate, archive, restore, delete and request a fresh run                |
| Answers               | Find answers by prompt, date, status, platform, source or sentiment; read full answer and citation evidence                         | Nothing; collected answers are evidence rather than editable content                                                                        |
| Groups                | Compare groups, inspect coverage and opportunities, analyse group visibility, sentiment, prompts, sources, competitors and evidence | Create and edit groups, change prompt membership, apply opportunities, and preview or manage grouping rules                                 |
| Brand and competitors | Read your own-brand identity, discover possible competitors, inspect aliases, compare performance and analyse evidence              | Confirm or configure your own brand, add or edit competitors, archive or restore them, manage aliases and apply reviewed identity decisions |
| Sources               | Explore cited domains, pages and supporting answers; export filtered source activity                                                | Classify a source domain, mark its tracking state, and save a target URL or notes                                                           |
| Tracked coverage      | Search published coverage, inspect its history and exact-page citation evidence, and measure citations and visibility over time     | Add published URLs, configure publication details, edit records, and archive or restore them                                                |
| Sentiment             | Analyse the overview, trends, platforms, brands, aspects and supporting quotations                                                  | Nothing; sentiment analysis is read-only                                                                                                    |
| AI referrals          | Analyse existing referral traffic, landing pages, referrers, devices, events and sessions                                           | Nothing; integrations cannot be configured through the connector                                                                            |
| Crawler logs          | Analyse crawler activity, pages, crawler identities and events; export filtered events                                              | Nothing; crawler integrations and credentials cannot be changed                                                                             |

The available actions depend on the permissions approved when you connect, your current Chatobserver role, your workspace plan and any features enabled for that workspace.

## What is not available yet

The connector cannot currently:

* create, edit, publish or download reports;
* manage alerts;
* create or configure data connections, AI referral integrations or crawler-log integrations;
* manage billing, subscriptions, workspace members or developer credentials;
* create or manage standalone personas; or
* buy a plan or take payment inside the assistant.

Tracked coverage is different from two similarly named capabilities:

* **Group coverage** describes how prompts are distributed across groups.
* **Source tracking** classifies a cited domain and can attach a target URL or notes to that domain.

Use [Tracked coverage](/docs/mcp/tracked-coverage) when you want to save a published article or PR URL and measure whether that exact page appears in collected answers.

## How to ask for work

Give the assistant the outcome and any important boundaries. For example:

* “In Clean Digital, show me the five prompts with the largest visibility decline over the last 30 days.”
* “Review this CSV against our existing prompts. Do not create anything yet.”
* “Create a group for the approved prompts, then show me exactly what will move before changing membership.”
* “Compare our own brand with active competitors and cite the answers behind the largest gap.”
* “Show negative sentiment themes and the source pages that appear in the supporting answers.”

For multi-step changes, ask the assistant to present a plan or preview first. Read-only research never needs to become a change just because the connector can make changes.

## Choose the next guide

* [Prompts and answers](/docs/mcp/prompts-and-answers)
* [Groups and grouping rules](/docs/mcp/groups)
* [Brands and competitors](/docs/mcp/competitors)
* [Sources and sentiment analysis](/docs/mcp/analysis)
* [Tracked coverage and PR measurement](/docs/mcp/tracked-coverage)
* [AI referrals and crawler activity](/docs/mcp/referrals-and-crawlers)
* [Usage, permissions and approvals](/docs/mcp/usage-and-safety)
* [Example requests and workflows](/docs/mcp/example-requests)