# Cursor Agent: Web Search, Browser and Site Access

> The Cursor agent can search the web, read links and control a browser. Cursor documents no user agent or robots.txt token for this traffic.

Cursor is an AI code editor. Its agent can search the web, read a link that a user gives it, and control a browser. Cursor documents no user agent, IP range or robots.txt token for this traffic, so a site owner cannot allow or block Cursor by name.

## Who makes Cursor

The company Anysphere builds Cursor. TechCrunch [reported this in October 2023](https://techcrunch.com/2023/10/11/anysphere-raises-8m-from-openai-to-build-an-ai-powered-ide), when Anysphere raised money from OpenAI. On August 14, 2026, Cursor wrote that it [is now a part of SpaceX](https://cursor.com/blog/joining-spacex) after SpaceX completed the acquisition.

## How the Cursor agent uses the web

The [agent tools page](https://cursor.com/docs/agent/tools) lists two tools that reach the web:

| Tool | What it does, according to Cursor |
|---|---|
| Web | Makes search queries and runs web searches. |
| Browser | Controls a browser to take screenshots, test applications and check visual changes. The agent can go to pages, use elements on them and record their current state. |

The [agent security page](https://cursor.com/docs/agent/security) says that the tools make network requests only to GitHub, to links for direct retrieval, and to web search providers. With the default configuration, agents cannot make other network requests. Cursor does not name its web search provider.

The [browser page](https://cursor.com/docs/agent/browser) says that the browser opens as a pane inside Cursor. The agent can go to any URL. Enterprise admins can limit the sites that the agent opens with a Browser Origin Allowlist.

## What a site owner can control

The Cursor pages above do not give these facts:

- A user agent string for searches, link retrieval or the browser.
- IP ranges or reverse DNS names.
- Whether the agent follows robots.txt.
- How the agent shows sources in its answers.

So no documented robots.txt group targets Cursor. A web search finds your pages only if the search provider has them in its index. The controls of that search engine apply, but Cursor does not say which engine it uses. If you see unwanted traffic from a tool, block it in your web server or firewall. Use the user agent and the IP addresses that your logs show.

---

[View the canonical page](https://aichatwatch.com/ai-chat-bot/cursor/) · [Browse llms.txt](https://aichatwatch.com/llms.txt)