# RankGrow — Full reference for AI agents

> Expanded reference for LLMs and coding agents. RankGrow exposes a remote MCP server at `https://mcp.rankgrow.io` (OAuth 2.1, streamable HTTP) with 35 tools for account context, websites, Google Search Console, SEO agent chats, and task workflows.

---

## What RankGrow does

RankGrow is an AI SEO platform that:
- Connects to Google Search Console to analyze real search performance
- Runs seven specialized agents (technical SEO, content, competitors, AI visibility, and more)
- Audits pages, surfaces quick wins, and produces prioritized fix lists
- Persists website profile memory (brand, audience, competitors, tone) across chats
- Converts recommendations into personal or website-scoped task backlogs

Public website: `https://www.rankgrow.io`
MCP docs: `https://www.rankgrow.io/mcp`

---

## MCP server

**Endpoint:** `https://mcp.rankgrow.io`
**Transport:** Streamable HTTP (MCP spec 2025-06-18+)
**Auth:** OAuth 2.1 (bearer token), discovery via `https://www.rankgrow.io/.well-known/oauth-protected-resource`
**Plan requirement:** Startup or Agency (MCP is not available on Free or Solo)

### Cursor (native HTTP — recommended)

Add to `.cursor/mcp.json` (project) or your global Cursor MCP config:

```json
{
  "mcpServers": {
    "rankgrow": {
      "url": "https://mcp.rankgrow.io"
    }
  }
}
```

On first use, Cursor shows **Needs login** — complete the RankGrow OAuth flow to authorize the session.

### Claude Code

```bash
claude mcp add --transport http rankgrow https://mcp.rankgrow.io
```

Restart Claude Code and run `/mcp` to authenticate.

### VS Code (GitHub Copilot)

```json
{
  "servers": {
    "rankgrow": {
      "type": "http",
      "url": "https://mcp.rankgrow.io"
    }
  }
}
```

Use **MCP: List Servers** → **RankGrow** → **Start Server**, then approve OAuth when prompted.

### ChatGPT

In **Settings → Connectors → Create** (Developer mode):
- **URL:** `https://mcp.rankgrow.io`
- **Authentication:** OAuth

### Codex CLI

```bash
codex mcp add rankgrow --url "https://mcp.rankgrow.io"
```

### OpenCode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "rankgrow": {
      "type": "remote",
      "url": "https://mcp.rankgrow.io",
      "enabled": true
    }
  }
}
```

### Discovery endpoints

- Server card: `https://www.rankgrow.io/.well-known/mcp/server-card.json`
- OAuth authorization server: `https://www.rankgrow.io/.well-known/oauth-authorization-server`
- OAuth protected resource: `https://www.rankgrow.io/.well-known/oauth-protected-resource`
- Curated index: `https://www.rankgrow.io/llms.txt`

Unauthenticated MCP requests receive `401` with `WWW-Authenticate: Bearer` so MCP-aware clients can auto-bootstrap OAuth.

---

## Tool catalog (35 tools)

### Account (5)
- `get_account_context` (read-only) — Returns the authenticated RankGrow user, account billing summary, and websites they can access.
- `list_websites` (read-only) — Lists RankGrow websites the authenticated user can access. Use websiteId from this result for website-scoped tools.
- `resolve_website` (read-only) — Finds accessible RankGrow websites by websiteId, name, slug, or domain.
- `get_plan_limits` (read-only) — Returns RankGrow plan limits and features. Defaults to the authenticated user's active account plan.
- `get_credit_usage` (read-only) — Returns account-level RankGrow credit usage for the authenticated user's billing account.

### SEO chats (9)
- `send_chat_message` — Sends a message to the RankGrow SEO agent. Provide websiteId for website-scoped chat, or omit it for personal chat.
- `list_chats` (read-only) — Lists personal chats or website-scoped RankGrow chats when websiteId is provided.
- `get_chat` (read-only) — Gets a single RankGrow chat by exact chatId.
- `get_chat_messages` (read-only) — Gets all persisted messages and parts for a RankGrow chat by exact chatId.
- `update_chat` — Updates RankGrow chat metadata such as name or pinned status.
- `delete_chat` — Deletes a RankGrow chat by exact chatId.
- `share_chat` — Creates or updates the public share for a RankGrow chat.
- `get_chat_share_status` (read-only) — Returns the current public share status for a RankGrow chat.
- `revoke_chat_share` — Revokes the public share for a RankGrow chat by exact chatId.

### Search Console & integrations (7)
- `get_search_console_connect_url` (read-only) — Returns the browser URL for connecting a Google Search Console account to RankGrow.
- `list_search_console_properties` (read-only) — Lists Google Search Console properties available through the user's connected Google accounts.
- `link_search_console_property` — Links or updates a website's Google Search Console property. Requires admin access to the website.
- `list_website_integrations` (read-only) — Lists integrations linked to a RankGrow website when the authenticated user can access it.
- `unlink_search_console_property` — Removes a website's Google Search Console integration. Requires admin access to the website.
- `list_connections` (read-only) — Lists the authenticated user's OAuth connections, optionally filtered by integration service.
- `delete_connection` — Deletes an OAuth connection owned by the authenticated user. Linked website integrations are removed by cascade.

### Tasks (7)
- `list_tasks` (read-only) — Lists personal tasks or website-scoped RankGrow tasks when websiteId is provided.
- `get_task` (read-only) — Gets a single personal or website-scoped RankGrow task by exact taskId.
- `create_task` — Creates a personal task or a website-scoped RankGrow task when websiteId is provided.
- `bulk_create_tasks` — Creates multiple personal tasks or website-scoped RankGrow tasks when websiteId is provided.
- `update_task` — Updates a personal or website-scoped RankGrow task by exact taskId.
- `delete_task` — Deletes a personal or website-scoped RankGrow task by exact taskId.
- `reorder_tasks` — Updates task status and position for personal or website-scoped RankGrow tasks.

### Websites (7)
- `get_website_context` (read-only) — Returns a RankGrow website and profile by websiteId when the authenticated user can access it.
- `create_website` — Creates a RankGrow website for the authenticated user from a domain or URL.
- `update_website` — Updates website metadata such as name, slug, or logo. Domain changes are not supported.
- `delete_website` — Deletes a RankGrow website by websiteId. Requires owner access and confirm=true.
- `get_website_profile` (read-only) — Returns the RankGrow website profile fields used for SEO and content context.
- `update_website_profile` — Creates or updates the RankGrow website profile fields used for SEO and content context.
- `discover_website_profile` — Discovers website profile fields from the website domain. Optionally saves the discovered profile.

---

## Operating notes for agents

- Call `get_account_context` or `list_websites` before website-scoped work. Use `resolve_website` when the user names a domain instead of a websiteId.
- Prefer `send_chat_message` for product-grade SEO analysis — it runs the full RankGrow agent pipeline with Search Console and website context instead of reasoning from scratch.
- Check `get_credit_usage` and `get_plan_limits` before large batched operations; MCP tool calls consume the same credits as the web app.
- Tools marked read-only in the catalog are safe to call speculatively; others mutate RankGrow state and require user intent.
- Google Search Console data is only available when the user has connected GSC in RankGrow and linked the property to the website.
- To avoid duplicate backlog noise, call `list_tasks` before `bulk_create_tasks`.
- Typical loop: resolve website → list integrations / Search Console property → `send_chat_message` with findings → `bulk_create_tasks` for fixes.

---

## Contact

- Website: https://www.rankgrow.io
- Contact form: https://www.rankgrow.io/contact
- MCP setup guide: https://www.rankgrow.io/mcp