# Hindsight Sets a New Standard: #1 on the BEAM Benchmark

![Graphic showing the Cursor logo above the Vectorize logo, with a plus sign between them, on a dark, abstract background with flowing purple and blue lines. Text reads “CURSOR + [vectorize]”](/content/_next/image?url=https%3A%2F%2Fmlrwd9rnffxq.i.optimole.com%2Fcb%3A641c.2be21%2Fw%3Aauto%2Fh%3Aauto%2Fq%3A90%2Ff%3Abest%2Fsm%3A0%2Fhttps%3A%2F%2Fblog.vectorize.io%2Fwp-content%2Fuploads%2F2025%2F09%2Fcursor-blog-post-1200x628-1.png&w=3840&q=75/index.html)

You’re midway through a PR review. Someone asks:

> _“What are the downstream dependencies of this service?”_

Normally, you’d open three browser tabs, search the wiki, and dig through architecture diagrams. By the time you find the answer, your review flow is broken.

With Cursor connected to a Vectorize agent:

- Ask the question
- The agent scopes docs by service metadata
- You get the exact diagram and context — right inside your IDE

## How It Fits Into Developer Work

You’re reviewing a pull request. A teammate flags a potential dependency issue, and instead of digging through diagrams, you pop open Cursor, type the question, and the service topology doc appears instantly — filtered to exactly the system in question.

Later that day, production throws a familiar error. Rather than combing through incident logs in a browser, you ask your agent for similar incidents in the last 30 days. The reports show up inline, scoped by severity, right where you’re already debugging.

When onboarding a new engineer, you no longer paste wiki links into Slack. They can just ask Cursor: _“How do I set up the API gateway locally?”_ and the correct runbook steps surface in their editor.

Your flow doesn’t break. The answers come to you.

## Quick Setup Recipe

Vectorize agents expose a structured search tool. Cursor connects to it via MCP and makes it available right in your command palette.

### Step 1: Create a Vectorize Agent

Make an MCP agent in the Vectorize UI or API. It hosts your tools and handles retrieval over your docs.

### Step 2: Add a Tool

Configure a tool and connect it to your pipeline. Set up parameters and link them to your pipeline’s metadata for automatic filtering.

### Step 3: Configure Cursor’s MCP settings

Open **Cursor Settings → MCP configuration** and add:

```json
{
  "mcpServers": {
    "vectorize-mcp": {
      "url": "https://agents.vectorize.io/api/agents/YOUR_AGENT_ID/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

This tells Cursor to connect directly to your Vectorize MCP agent over HTTPS.

### Step 4: Use It in Cursor

Restart Cursor. Your Vectorize agent will now appear in the Agent panel (`⌘I`), accessible from the command palette.

**Important:** Cursor’s AI won’t automatically decide to use your MCP tool. You need to explicitly ask it.

## Get Started

1. **[Sign up for Vectorize](https://platform.vectorize.io/)**
2. **[Install Cursor](https://cursor.sh/)**
3. **[Follow the integration guide](https://docs.vectorize.io/build-deploy/agents/mcp/integrations/cursor)**
4. Add your agent in Cursor Settings, restart Cursor, and start asking!

[Turn Messy Docs into Agent-Ready Data. Use Vectorize Iris to extract structured content from scanned PDFs, contracts, and more — no manual cleanup.\ Try Vectorize Iris →](https://platform.vectorize.io/)