Overview

The Kwrds.ai MCP Server implements a Model Context Protocol (MCP) compatible backend that exposes keyword research, People Also Ask (PAA), SERP snapshot and related SEO utilities as machine-friendly endpoints. It is designed to power integrations (chatbots, agents, dashboards) that need programmatic access to search intelligence outputs without embedding scraping logic or SERP processing in the client.

For developers this server is useful because it centralizes SERP parsing, keyword discovery and enrichment into a single service that can be queried by MCP-aware models or standard HTTP clients. That lets you focus on product UX or agent logic while delegating keyword and SERP operations to a repeatable, versioned service.

Features

Keyword discovery and long-tail suggestion generation
People Also Ask (PAA) extraction and question clustering
SERP snapshot: organic results, featured snippets, knowledge panels, and other SERP features
Competitor keyword overlap and gap analysis helpers
API-compatible with MCP-style request/response patterns (JSON)
Docker-friendly and configurable via environment variables
Designed to be integrated with language models and automation pipelines

Installation / Configuration

Prerequisites:

Node.js 16+ (or another runtime depending on the repo; examples below use Node)
Docker (optional)
API keys for any upstream services you plan to use (Google Custom Search, SerpAPI, OpenAI, etc.)

Quickstart (Git + Node):

# clone
git clone https://github.com/mkotsollaris/kwrds_ai_mcp.git
cd kwrds_ai_mcp

# install
npm install

# copy env and edit
cp .env.example .env
# edit .env to add API keys and change PORT (e.g. PORT=3000)

# run dev server
npm run dev
# or production
npm run start

Docker:

# build
docker build -t kwrds-mcp .

# run (map port and pass environment vars)
docker run -p 3000:3000 \
  -e PORT=3000 \
  -e SERP_API_KEY=your_key_here \
  -e OTHER_KEY=... \
  kwrds-mcp

Configuration notes:

.env keys typically include SERP API provider credentials, optional LLM keys for enrichment, and a PORT.
Look for a config file or environment references in the repo to customize provider selection, request rate limits, and cache options.

Available Tools / Resources

The MCP server exposes a set of logical tools/endpoints useful for SEO workflows. Typical resources include:

Tool	Purpose
keyword_research	Expand a seed keyword into related suggestions, search volumes, difficulty signals
people_also_ask	Retrieve PAA questions and metadata for a query
serp_snapshot	Return parsed SERP components for a query (rankings, snippets, features)
competitor_keywords	Compare keyword sets between domains and surface gaps
trends	Surface trending keywords or rising queries for a topic or country

Endpoints are generally accessed via JSON POSTs shaped for MCP consumers; the server responds with structured JSON that contains both the raw data and an optional natural-language summary for model consumption.

Example request (curl):

curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -d '{
    "tool":"keyword_research",
    "input":{"query":"vegan protein powder","country":"US","limit":20},
    "context": {"user":"seo_bot"}
  }'

Example simplified response:

{
  "tool":"keyword_research",
  "query":"vegan protein powder",
  "results":[
    {"term":"best vegan protein powder","volume":54000,"intent":"commercial"},
    {"term":"vegan protein powder for athletes","volume":2400,"intent":"informational"}
  ],
  "summary":"Top long-tail terms include purchase-intent phrases and athlete-focused queries."
}

Use Cases

Content planning: Generate prioritized topic clusters and long-tail keywords from a seed phrase (e.g., expand “home espresso machine” into blog-topic ideas and average volumes).
Question-driven content: Use People Also Ask extraction to compile FAQ sections or to generate structured Q&A content for a landing page (e.g., fetch PAA for “how to start podcast” and generate answers).
SERP monitoring and alerts: Periodically snapshot a set of target queries and detect changes in SERP features (featured snippet gained/lost, new knowledge panel).
Competitor analysis: Compare the top domains for a money keyword and compute missed keyword opportunities to guide outreach or content updates.
Agent augmentation: Feed MCP responses into an LLM-based assistant so the model can cite up-to-date SERP facts, propose title/meta updates, or suggest target keywords.

Tips for Developers

Cache results for common queries to reduce upstream API costs and improve latency.
Normalize country and language codes in requests to get consistent SERP outputs.
When integrating with LLMs, pass both the raw structured output and a short natural-language summary so models can reason and generate human-facing content.
Use Docker for reproducible deployments and to encapsulate any scraping dependencies.

If you need more granular integration examples (client SDK snippets, schema documentation, or deployment templates), consult the repository docs and example scripts in the project.

Kwrds.ai MCP Server Keyword Research & SERP Tools