---
title: "Cursor custom docs + LLMind: rich file context for AI coding | LLMind"
description: "Point Cursor custom docs at a directory of LLMind-enriched files. The AI coding assistant reads structured metadata alongside raw file content — no vector DB."
url: https://llmind.org/mcp/cursor/
source_format: html
---
# Cursor custom docs with LLMind-enriched files

Published 2026-04-22 · 6 min read

Cursor is an AI-first code editor. It supports both MCP servers and its own custom-docs feature — both of which can read LLMind-enriched files. This guide covers both integration paths so you can pick whichever fits your workflow.

## Two paths: MCP or custom docs

Cursor offers two ways to give the AI coding assistant context about your files:

-   **MCP (standard):** Connect via the filesystem MCP server. Same pattern as Claude Desktop and Windsurf. Works across tools.
-   **Cursor custom docs (Cursor-native):** Point Cursor at a documentation directory. Cursor indexes it and makes it available to the AI. More integrated into the Cursor editing experience.

Both work with LLMind-enriched files. Either one is fine; pick whichever feels more natural for your workflow.

## Path 1: MCP

The MCP approach is the same as Claude Desktop: enrich a directory with LLMind, then configure the filesystem MCP server in Cursor.

### Step 1 — Install LLMind

```
pipx install 'llmind-cli[all]'
```

### Step 2 — Enrich your directory

```
llmind enrich --recursive ~/docs/
```

### Step 3 — Configure Cursor MCP

Open Cursor Settings. Navigate to Features → MCP. Add a new MCP server:

-   **Name:** docs (or whatever you want to call it)
-   **Command:** npx
-   **Arguments:** `-y`, `@modelcontextprotocol/server-filesystem`, `/Users/you/docs`

Replace `/Users/you/docs` with your enriched directory path. Save and restart Cursor.

### Step 4 — Test in Cursor

Open a new file or start a new chat in Cursor. Type `@Files` or `@Docs` in the prompt to reference files from the MCP server. Cursor will read the file and its embedded semantic metadata.

## Path 2: Cursor custom docs

Cursor's custom-docs feature indexes a directory of documentation. It's simpler to set up than MCP if you're not using MCP elsewhere.

### Step 1 — Install LLMind

```
pipx install 'llmind-cli[all]'
```

### Step 2 — Enrich your directory

```
llmind enrich --recursive ~/docs/
```

### Step 3 — Add custom docs to Cursor

Open Cursor Settings. Navigate to Features → Docs. Click "Add custom docs". Select the directory you enriched in Step 2. Cursor will index it and make the files available to the AI.

### Step 4 — Use @Docs in your prompts

In any Cursor chat or prompt, type `@Docs` to reference your documentation. Cursor will include the files and their metadata in the context.

## Combining both paths

You don't have to choose. You can set up both MCP and custom docs pointing at the same enriched directory. Use whichever is more convenient in the moment.

## Troubleshooting

**MCP server not appearing in Cursor:** Restart Cursor completely after adding the MCP config. Check Cursor's settings UI to verify the entry was saved.

**Custom docs not indexing:** Make sure the directory exists and Cursor has read access. Try selecting the directory again in Settings → Features → Docs.

**Files not showing up:** Verify LLMind enrichment completed: `llmind enrich --recursive ~/docs/` should have run without errors. Check that the directory path in your Cursor config matches the enriched directory.

## Related

-   [MCP integration hub](https://llmind.org/mcp/)
-   [Claude Desktop integration](https://llmind.org/mcp/claude-desktop/)
-   [Give Claude access to your files](https://llmind.org/use-cases/ai-agents-file-access/)
-   [Product overview](https://llmind.org/product/)
-   [For you](https://llmind.org/for/)
-   [Learning resources](https://llmind.org/learn/)
-   [Documentation](https://llmind.org/docs/)

[Star on GitHub](https://github.com/dmitryrollins/LLMind) · [Install CLI](https://llmind.org/install/)