arenza-mcp-client

TypeScript client for the Arenza MCP server — programmatic access to AI visibility metrics, brand mentions, hallucination findings, and GEO opportunities across ChatGPT, Claude, Gemini, Perplexity, Copilot, and Grok.

Arenza is a Generative Engine Optimization (GEO) platform that measures how 6 leading AI assistants describe brands across 4 markets (US, UK, DE, JP). This client wraps the Model Context Protocol server at mcp.arenza.ai, giving you typed access to the same data the Arenza dashboard renders.

Why an MCP client (not just a REST SDK)

The Arenza data surface ships as both a REST API (api.arenza.ai) and an MCP server (mcp.arenza.ai). The MCP variant is what AI agents — Claude Desktop, Claude Code, Cursor, Continue, and any custom agent built on the MCP protocol — connect to natively. This client lets you embed the same connection inside a TypeScript backend, automation, or chatbot.

If you want to use Arenza from Claude Desktop directly (no code), see Use Claude with the Arenza MCP server instead.

Install

npm install @arenza/mcp-client
# or
pnpm add @arenza/mcp-client
# or
yarn add @arenza/mcp-client

Quick start

import { ArenzaMCPClient } from '@arenza/mcp-client';

const client = new ArenzaMCPClient({
  // Get a token at https://app.arenza.ai/settings/api or use OAuth flow.
  token: process.env.ARENZA_TOKEN!,
});

// 10 read+write tools available — see tools list below.
const brands = await client.listBrands();
console.log(brands);

const overview = await client.getBrandOverview({ brand_id: brands[0].id });
console.log(overview.share_of_voice, overview.wrong_claims);

Tools

The Arenza MCP server exposes 10 tools:

Read

| Tool | Description | |---|---| | list_brands | List all brands in the authenticated tenant's portfolio. | | get_brand_overview | Aggregate visibility + accuracy snapshot for one brand. | | list_prompts | List the AI prompts probed for a brand, with mention rates per LLM. | | list_opportunities | List measurement-led GEO opportunities (wrong claims to fix, missing canonical pages, listicle gaps). | | suggest_competitors | LLM-suggested competitors based on the brand description. | | suggest_prompts | LLM-generated buyer-perspective prompts (70%+ unbranded ratio enforced). |

Write

| Tool | Description | |---|---| | add_competitor | Add a competitor to a brand's tracking list. | | dismiss_competitor | Remove a competitor (e.g. wrong suggestion). | | mark_opportunity_done | Mark a GEO opportunity as completed. | | generate_geo_article | Draft a canonical-fact article body anchored to a specific finding. |

Each tool is fully typed in this client — TypeScript autocomplete shows parameter shapes and return types.

Authentication

Two options:

API token (simplest for backend / cron jobs):

const client = new ArenzaMCPClient({ token: process.env.ARENZA_TOKEN! });

Get one at app.arenza.ai/settings/api.

OAuth (for multi-tenant apps, AI agents, end-user authorization):

const client = await ArenzaMCPClient.fromOAuth({
  clientId: 'your-client-id',
  redirectUri: 'https://yourapp.com/oauth/callback',
});

The Arenza MCP server supports OAuth 2.0 with Dynamic Client Registration (DCR) and PKCE — full spec at mcp.arenza.ai/.well-known/oauth-authorization-server.

Example: weekly GEO digest cron

import { ArenzaMCPClient } from '@arenza/mcp-client';

async function weeklyDigest() {
  const client = new ArenzaMCPClient({ token: process.env.ARENZA_TOKEN! });
  const brands = await client.listBrands();

  for (const brand of brands) {
    const ovw = await client.getBrandOverview({ brand_id: brand.id });
    const opps = await client.listOpportunities({ brand_id: brand.id });
    console.log(
      `${brand.name}: SoV ${ovw.share_of_voice}%, ${ovw.wrong_claims} wrong claims, ${opps.length} opportunities`,
    );
  }
}

weeklyDigest();

Rate limits

• Free tier: 100 MCP calls/hour
• Pro tier ($299/mo): 1,000 calls/hour
• Protect tier ($799/mo): 10,000 calls/hour
• Enterprise: unlimited

Current limits are returned in X-RateLimit-Remaining headers. The client does not auto-retry on 429 — retry yourself with backoff.

Related projects

• @arenza/cli — npx arenza scan brand.com for one-off scans.
• arenza-mcp-client-python — Python equivalent of this client.
• @arenza/langchain — wrap Arenza tools as LangChain tools.
• @arenza/llamaindex — same for LlamaIndex.
• @arenza/vercel-ai-sdk — Vercel AI SDK provider.
• awesome-geo — broader curated GEO resources list.

Resources

• Arenza homepage: https://arenza.ai
• API contract: https://app.arenza.ai/methodology
• Long-form guides: https://arenza.ai/guides
• Brand reference for AI assistants: https://arenza.ai/llms.txt + https://arenza.ai/llms-full.txt
• MCP server: https://mcp.arenza.ai
• OAuth spec: https://mcp.arenza.ai/.well-known/oauth-authorization-server

License

MIT © Arenza