How to connect Cursor to your CapturedIt library via MCP

Cursor is where you write code. Your CapturedIt library is where you save research, API docs, architecture decisions, and bug investigation notes. Connecting them via MCP means Cursor can search and reference your library without you copy-pasting context into every conversation.

Why developers connect Cursor to a knowledge base

• Reference saved API documentation without leaving the editor
• Pull in architecture notes when implementing a feature
• Search past bug investigations when a similar error appears
• Save new findings back to the library from within Cursor

For a broader developer workflow, see CapturedIt for developers.

Prerequisites

• CapturedIt Vault plan (MCP access requires Vault)
• Cursor with MCP support enabled
• A CapturedIt API key with appropriate scopes

Setup steps

1. Generate an API key: In CapturedIt, go to Dashboard → Connections → API. Create a key with read and write scopes if you want Cursor to save captures back to your library.
2. Open Cursor MCP settings: Add a new MCP server pointing to CapturedIt's MCP endpoint. Configuration details are in Dashboard → Integrations → Cursor, including a one-click setup option where available.
3. Authenticate: Provide your API key in the MCP server configuration.
4. Verify: Ask Cursor to search your CapturedIt library for a known item. If it returns results, the connection is working.

What Cursor can do with your library

• Search items by keyword or concept
• Retrieve full content from specific captures
• Create new captures (with write-scoped keys)
• Answer questions grounded in your saved research

Private items remain excluded. Only captures you have marked as accessible are returned.

MCP vs Context Hub REST API

MCP is the right integration for Cursor because it speaks Cursor's native tool protocol. For custom scripts, CI pipelines, or non-MCP clients, use the Context Hub REST API instead. Vault includes both.

New to MCP? Read what is MCP for background on the protocol.