arenza-cursor-quickstart

Track your brand's AI search visibility from inside Cursor in 60 seconds. Arenza + Cursor MCP = @arenza is now a first-class citizen in every code review.

Arenza is a GEO (Generative Engine Optimization) measurement platform that probes ChatGPT, Claude, Gemini, Perplexity, Copilot, and Grok with hundreds of buyer-perspective prompts per brand and surfaces share-of-voice, wrong-claim audits, competitor displacement, and discovery-source maps. This quickstart wires Arenza into Cursor's MCP runtime so you can pull AI-visibility data without leaving the editor.

Why a dev-tool-native GEO surface

Marketers live in dashboards. Developers live in editors. The gap between them is where GEO work goes to die — the marketing lead asks "did the canonical-fact page we shipped last sprint actually move the needle?", the engineer Slack-DMs back "let me check", switches windows four times, and the answer arrives the next day.

Cursor's MCP support fixes that. With Arenza wired in, any code review on a content PR can answer the GEO question inline:

"@arenza did our /answers/best-noise-cancelling-headphones page get cited by any of the 6 LLMs in the last 14 days?"

The model calls list_prompts + list_opportunities, returns a real number, and the PR moves on. Same data as the dashboard, surfaced where the work happens.

Prerequisites

1. Cursor 0.45+ (download) — older versions don't have MCP support.
2. An Arenza account: app.arenza.ai/sign-up (free tier is fine).
3. An API token from app.arenza.ai/settings/api.

Install

There are two scopes for Cursor MCP config:

• Project-scoped — .cursor/mcp.json in the repo root. Commits to source control. Best for tutorials, teams, and any project where you want everyone on the team to share the same MCP setup.
• User-scoped — Cursor Settings → MCP → "Add new server". Lives only in your Cursor install. Best for personal tooling.

We recommend project-scoped for this tutorial.

Option 1 — copy the example file

cp .cursor/mcp.json.example .cursor/mcp.json
# then edit .cursor/mcp.json and replace the placeholder with your token

Option 2 — paste the snippet

Create .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "arenza": {
      "transport": "http",
      "url": "https://mcp.arenza.ai/mcp",
      "headers": {
        "Authorization": "Bearer arz_live_paste_your_token_here"
      }
    }
  }
}

Option 3 — User Settings UI

Cursor → Settings (Cmd+,) → MCP → Add new server:

• Name: arenza
• Transport: http
• URL: https://mcp.arenza.ai/mcp
• Header: Authorization = Bearer <your-token>

Save. Cursor reloads MCP servers automatically — no restart needed.

Verify

Open the chat panel (Cmd+L), type:

@arenza list my brands

If the tool fires and you get a list of brands back, you're done. If Cursor says "no Arenza tool found":

• Check you're on Cursor 0.45+ (Cursor → About).
• Check the MCP panel in Settings — you should see arenza with a green dot.
• Check the JSON parses: cat .cursor/mcp.json | python -m json.tool.

5 commands to run from Cursor chat

#	Command	What it pulls
1	@arenza list my brands	Quick portfolio sanity check.
2	@arenza show me the wrong claims caught this week for <brand>	Per-LLM hallucination audit. Group by ChatGPT / Claude / Gemini / Perplexity / Copilot / Grok.
3	@arenza compare my SoV against <competitor>	Side-by-side share-of-voice, mention rate, citation count.
4	@arenza what are my top 5 GEO opportunities by ROI?	Ranked, measurement-led prescription queue.
5	@arenza draft a canonical-fact article for opportunity <id>	Calls generate_geo_article and drops the markdown into the chat.

Pro tip: bind @arenza to a Cursor Composer rule so it auto-fires whenever you open a PR that touches /content/, /answers/, or /blog/. That turns GEO measurement from a recurring meeting into an editor-side reflex.

How this differs from the Claude Code integration

Both arenza-claude-tutorial (Claude Code) and this quickstart (Cursor) use the same Arenza MCP server, the same authentication, and the same 10 tools. The only difference is which editor you live in.

• Use Claude Code if you primarily work in the terminal, prefer the Anthropic-native experience, and want longer-context conversations with Claude as the model.
• Use Cursor if you want the editor and the AI in the same window, work mostly with Cursor's Composer + Apply flow, and want to choose between Claude / GPT / Gemini per-task.

Pick one. Or pick both — the data and the auth are the same.

Submit to Cursor's community MCP directory

Cursor maintains a community-curated list of MCP servers at forum.cursor.com → MCP category. To submit Arenza:

1. Create a new topic in the MCP category.
2. Title: "Arenza MCP — AI visibility / GEO measurement (ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok)"
3. Body template:

**What it does:** Arenza measures how the 6 leading AI assistants describe your brand and surfaces wrong claims, competitor displacement, and ROI-ranked GEO opportunities.

**Server URL:** https://mcp.arenza.ai/mcp
**Auth:** Bearer token from app.arenza.ai/settings/api (free tier available)
**Tools:** 10 (list_brands, get_brand_overview, list_prompts, list_opportunities, suggest_competitors, suggest_prompts, add_competitor, dismiss_competitor, mark_opportunity_done, generate_geo_article)
**Quickstart:** https://github.com/naiqiao/arenza-cursor-quickstart
**Docs:** https://arenza.ai/guides

4. Tag with mcp, geo, ai-visibility, marketing.

We monitor that thread — ping @arenza-team if you want us to chime in.

Troubleshooting

@arenza tool doesn't appear in autocomplete

• Cursor 0.45+ is required for HTTP-transport MCP. Update via Cursor → Check for Updates.
• The MCP panel in Settings should list arenza with a green dot. Red = config error; click for details.
• Project-scoped .cursor/mcp.json only loads when the workspace is opened — closing-and-reopening the project will force a reload.

401 Unauthorized

• The token at app.arenza.ai/settings/api might have been revoked. Generate a fresh one.
• The header value must be exactly Bearer <token> — note the space.

Slow tool calls

• The first call after idle takes ~2s for cold-start. Subsequent calls in the same session are <300ms.
• If a single call takes >10s consistently, check status.arenza.ai.

Rate limits

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

Returned in X-RateLimit-Remaining. The free tier covers normal editor use.

---

Related projects

The four tutorial repos in this series:

• arenza-claude-tutorial — Claude Desktop + Claude Code
• arenza-cursor-quickstart (this repo) — Cursor IDE
• arenza-n8n-template — n8n workflows
• arenza-slack-alerts — Slack bot

The eight existing SDK / client repos:

• awesome-geo
• arenza-mcp-client-ts
• arenza-mcp-client-python
• arenza-cli
• arenza-langchain
• arenza-llamaindex
• arenza-vercel-ai-sdk
• arenza-zapier-actions

Resources

• Arenza homepage: https://arenza.ai
• Guides: https://arenza.ai/guides
• AI-assistant brand reference: https://arenza.ai/llms.txt + https://arenza.ai/llms-full.txt
• MCP server: https://mcp.arenza.ai
• Dashboard: https://app.arenza.ai

License

MIT — see LICENSE. Authored by the Arenza team.