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

# MCP overview

> Use Claude and other AI assistants to explore your Hikoo data through the Model Context Protocol.

Hikoo runs a hosted **Model Context Protocol (MCP) server** so you can ask Claude, or any MCP-compatible client, about your GEO data in plain language. Citations, audits, prompts, competitors, source intelligence, and more, without leaving your chat.

Ask things like:

* "How is my site cited compared to competitors this month?"
* "List the pages that failed criteria in my latest audit."
* "Add a prompt about pricing transparency under the pricing topic."

Your client calls the Hikoo MCP server behind the scenes, which reads and writes data in your Hikoo workspaces.

## How it works

<Columns cols={3}>
  <Card title="Hosted" icon="cloud">
    Hikoo runs the server. You do not install anything locally. Point your client at our URL.
  </Card>

  <Card title="One key, your account" icon="key">
    A single API key identifies you as a user. It can reach every website in the workspaces you belong to.
  </Card>

  <Card title="Read and write" icon="wand-sparkles">
    Read citations, audits, and analytics. Create or update prompts and competitors. Queue new audits.
  </Card>
</Columns>

<Note>
  The key is scoped to **your user account**, not to a single website. Your client first discovers which websites you have, then passes the chosen `website_id` to each data tool. This replaces the older per-website key model.
</Note>

## What you get

The server ships **70 tools** across these areas:

* **Citations and sources.** Visibility trends, top platforms, share of voice, top source domains, and citation breakdowns.
* **Sentiment.** Summary and trend of how AI answers portray your brand (available on some plans).
* **Prompts and topics.** Create, update, batch-update, and delete prompts; organize them into topics; read per-prompt stats and citations.
* **Competitors.** Share-of-voice ranking, per-competitor citations, suggestions, and add, update, or remove competitors.
* **Audits.** List past audits, inspect page-level criteria, and queue full-site, single-URL, or selected-page audits.
* **Health.** robots.txt, llms.txt, and sitemap status, plus improvement suggestions.
* **Indexation and pages.** Indexed-page inventory, summaries, common errors, and re-crawls.
* **Content gap.** Suggestions where competitors are cited and you are not.
* **Source intelligence.** AI-cited domains ranked by frequency, so an agent can automate listing claims and outreach.
* **Discovery and subscription.** Who you are, your workspaces and websites, and plan details.

## Who should use it

* **Power users** comfortable in Claude Desktop, Claude Code, or any MCP client.
* **Analysts** who prefer ad-hoc questions over dashboards.
* **Teams** building AI workflows that need programmatic access to their GEO data.

## Next steps

<Columns cols={2}>
  <Card title="Set up MCP" icon="plug" href="/guides/mcp/setup">
    Generate an API key and connect your AI client.
  </Card>

  <Card title="Tools reference" icon="wrench" href="/guides/mcp/tools">
    Every tool the MCP exposes, grouped by area.
  </Card>
</Columns>