Skip to main content

​
What You Can Do

No need to upload documents to Dify. By connecting Linkly AI via MCP, let Dify’s Agent use the search tool to retrieve documents stored on your computer in real time — quickly build an internal knowledge Q&A assistant for your team.
Unlike Dify’s built-in vector knowledge base, Linkly AI reads local files directly. Documents are updated without needing to re-index, and the Agent retrieves the latest content every time — always in sync with your computer.
Use Dify’s visual interface to chain together Linkly AI’s search (retrieval), outline (outline), and read (close reading) tools to quickly build well-structured document summary, Q&A, or content extraction application prototypes.

​
Prerequisites

  • Dify installed and running
  • Linkly AI Desktop running with the MCP service active

​
Configuration Steps

Choose the appropriate access mode based on where your Dify is deployed:
When Dify and Linkly AI are running on the same computer, use the local address directly:Add an MCP tool in Dify and configure:
  • URL: http://127.0.0.1:60606/mcp
  • No additional authentication required
If Dify is running in a Docker container, 127.0.0.1 points to the container itself. You need to use the host machine IP or host.docker.internal to access the local Linkly AI.

​
Application Examples

Once configured, you can build various AI applications in Dify using the document retrieval capabilities provided by Linkly AI, for example:
  • A Q&A bot based on private documents
  • A document summary generator
  • An intelligent document search assistant

​
FAQ

Check based on your deployment method: - Same computer: Confirm that the Linkly AI MCP service is running. If Dify is in Docker, use host.docker.internal instead of 127.0.0.1 - LAN: Confirm that LAN access is enabled and the access token is correct - Cloud: Confirm the tunnel is connected and the API Key is correct