Rapid Indexer
Send new pages and backlinks to Googlebot fast, then check which URLs Google has indexed.
Install
Add this to your client's MCP configuration (Claude Desktop: claude_desktop_config.json; Cursor: .cursor/mcp.json; VS Code and others use the same "mcpServers" shape).
claude mcp add --transport http rapid-indexer 'https://mcp.rapid-indexer.com/mcp' --header 'Authorization: <Bearer <your Rapid Indexer API key> (get one at https://rapid-indexer.com/api_access)>'
{
"mcpServers": {
"rapid-indexer": {
"type": "http",
"url": "https://mcp.rapid-indexer.com/mcp",
"headers": {
"Authorization": "<Bearer <your Rapid Indexer API key> (get one at https://rapid-indexer.com/api_access)>"
}
}
}
}
Documentation
From the project's README on GitHub, refreshed weekly. Links open on GitHub.
Contents
rapid-indexer-mcp
Model Context Protocol server for Rapid Indexer. It lets Claude, Cursor, ChatGPT, Claude Code and any other MCP host submit URLs to Googlebot, check whether pages are indexed, read task results and launch traffic / CTR campaigns using your Rapid Indexer API key.
Built on the official TypeScript SDK v2 (@modelcontextprotocol/server). Serves over stdio (local, one account) or Streamable HTTP (hosted, multi-tenant: each request carries its own API key).
Requirements
- Node.js 20+
- A Rapid Indexer API key: log in, open your account menu → API Access (
https://rapid-indexer.com/api_access).
Quick start (stdio)
Cursor
~/.cursor/mcp.json (or .cursor/mcp.json in a project):
{
"mcpServers": {
"rapid-indexer": {
"command": "npx",
"args": ["-y", "rapid-indexer-mcp"],
"env": { "RAPID_INDEXER_API_KEY": "YOUR_API_KEY" }
}
}
}
Claude Desktop
~/.config/Claude/claude_desktop_config.json on Linux:
{
"mcpServers": {
"rapid-indexer": {
"command": "npx",
"args": ["-y", "rapid-indexer-mcp"],
"env": { "RAPID_INDEXER_API_KEY": "YOUR_API_KEY" }
}
}
}
Claude Code
claude mcp add rapid-indexer -e RAPID_INDEXER_API_KEY=YOUR_API_KEY -- npx -y rapid-indexer-mcp
From a local checkout (before publishing to npm)
cd mcp
npm install
npm run build
Then point the host at the built CLI:
{
"mcpServers": {
"rapid-indexer": {
"command": "node",
"args": ["/absolute/path/to/Rapid-Indexer-main/mcp/dist/cli.js"],
"env": { "RAPID_INDEXER_API_KEY": "YOUR_API_KEY" }
}
}
}
Hosted mode (Streamable HTTP)
One process serves every customer; the API key travels with each request.
RAPID_INDEXER_BASE_URL=https://rapid-indexer.com \
node dist/cli.js --http --host 0.0.0.0 --port 3333 --path /mcp
Clients connect to https://your-host/mcp and send Authorization: Bearer <api key> (also accepted: X-API-Key: <key>, or ?api_key= on the URL). GET /healthz returns a liveness JSON.
Cursor example for a remote server:
{
"mcpServers": {
"rapid-indexer": {
"url": "https://mcp.rapid-indexer.com/mcp",
"headers": { "Authorization": "Bearer YOUR_API_KEY" }
}
}
}
Put it behind TLS (nginx/Caddy) and run it with systemd or pm2. The server is stateless, so you can run several replicas. --api-key / RAPID_INDEXER_API_KEY in HTTP mode is only a fallback for requests that carry no key (single-tenant self-hosting).
Tools
| Tool | Spends credits | What it does |
|---|---|---|
get_account | no | Account id, email, credit balance (+ USD). |
get_pricing | no | Live credits per URL (standard / VIP / checker), traffic price per 1,000 visitors, USD per credit. |
estimate_cost | no | Cost of N URLs as indexer/checker (+VIP) and whether the balance covers it. |
submit_urls_for_indexing | yes | Create an indexing task (standard or VIP queue, optional drip feed). Max 10,000 URLs. |
check_index_status | yes (~0.1 cr/URL) | Create an index-check task: per-URL indexed / unindexed. |
list_tasks | no | Paginated task list with indexed/unindexed/pending/error counts; filter by type/status. |
get_task | no | Refresh and return one task's status and progress. |
get_task_links | no | Per-URL results with status filter and pagination. |
create_traffic_campaign | yes | AI visitors to a URL (geo, device, referrer, behaviour). |
create_ctr_campaign | yes | Visitors search keywords on Google/Bing/… and click your result. |
list_search_engines | no | Valid search_engine values for CTR campaigns. |
check_service_health | no | API version, enabled features, and confirms the key works. |
Every tool returns human-readable text plus structuredContent validated by an output schema. Errors from the API come back as isError results with the API message and a hint (top-up link on insufficient credits, key location on 401).
Prompt: index_and_verify — guided estimate → submit → verify workflow.
Environment / flags
| Variable | Flag | Default | Notes |
|---|---|---|---|
RAPID_INDEXER_API_KEY | --api-key | – | Required for stdio. Fallback only in HTTP mode. |
RAPID_INDEXER_BASE_URL | --base-url | https://rapid-indexer.com/api/v1/index.php | Site origin is also accepted; the router path is appended. |
HOST | --host | 127.0.0.1 | HTTP mode. |
PORT | --port | 3333 | HTTP mode. |
MCP_PATH | --path | /mcp | HTTP mode. |
Development
npm install
npm run build # tsc → dist/
npm run dev # run src/cli.ts with tsx (stdio)
npm run inspect # MCP Inspector against dist/cli.js
If npm install hangs on a machine with broken IPv6, run it with
NODE_OPTIONS="--no-network-family-autoselection --dns-result-order=ipv4first".
Publishing
npm version patch
npm publish --access public
prepublishOnly builds dist/. The package ships only dist/, README.md and LICENSE.
API coverage
The server talks to the public REST API documented at https://rapid-indexer.com/api-docs. list_tasks and pricing were added to that API together with this server; hosts running an older build of the site will see those two tools fail with "Invalid action" until the API is redeployed.
License
MIT
Frequently asked questions
What is Rapid Indexer?
Send new pages and backlinks to Googlebot fast, then check which URLs Google has indexed.
How do I install Rapid Indexer?
It runs as a remote server. In Claude Code: claude mcp add --transport http rapid-indexer 'https://mcp.rapid-indexer.com/mcp' --header 'Authorization: <Bearer <your Rapid Indexer API key> (get one at https://rapid-indexer.com/api_access)>'. For other clients, paste the JSON configuration from this page into their MCP settings.
Does Rapid Indexer need an API key?
Yes. It expects Authorization, which you set in its configuration.
Is Rapid Indexer open source?
Yes. Its code is on GitHub under the MIT license.
Who maintains Rapid Indexer?
It is published by rollounden. AI.tn lists it but does not maintain or endorse it.
From the official MCP Registry. Listing is not an endorsement: review a server's code and permissions before giving it access to your data.