> ## Documentation Index
> Fetch the complete documentation index at: https://linkly.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Using Linkly AI in Claude Cowork

> Integrate your Linkly AI local knowledge base with Claude Cowork, enabling the AI desktop assistant to search and read your private documents

## What You Can Do

<AccordionGroup>
  <Accordion title="Local direct connection — no tunnel needed, high-frequency usage anytime">
    Unlike the claude.ai web version, Claude Desktop connects to Linkly AI
    directly via local MCP — no remote tunnel needed, lower latency. Ask "What
    have I recorded about XX?" during daily work, and Linkly AI's `search` tool
    instantly searches all your notes on your computer for a faster response.
  </Accordion>

  <Accordion title="Deep analysis of local long documents">
    Claude excels at processing very long text. Use Linkly AI's `outline` tool
    to first view the table of contents of a manuscript or report, then use
    `read` to carefully read key chapters — far more efficient than pasting the
    entire text, and without hitting context length limits.
  </Accordion>

  <Accordion title="Collaborate with other local MCP servers">
    Claude Desktop supports connecting to multiple MCP servers simultaneously.
    Linkly AI can work together with filesystem MCP, database MCP, and more,
    allowing Claude to access your knowledge base, local files, and other data
    sources all in one conversation, completing more complex tasks.
  </Accordion>
</AccordionGroup>

## Prerequisites

* [Claude Desktop](https://claude.ai/download) installed (Claude Cowork is Claude Desktop; requires Claude Pro or higher subscription)
* Linkly AI Desktop running with the MCP service active

## Configuration Steps

<Tabs>
  <Tab title="MCP HTTP (Recommended)" icon="globe">
    <Steps>
      <Step title="Open the configuration file">
        In the Claude menu, select **Settings → Developer → Edit Config** to open
        `claude_desktop_config.json`.
      </Step>

      <Step title="Add Linkly AI MCP">
        Add the following configuration to the `mcpServers` object:

        ```json theme={null}
        {
          "mcpServers": {
            "linkly-ai": {
              "url": "http://127.0.0.1:60606/mcp"
            }
          }
        }
        ```
      </Step>

      <Step title="Restart Claude">
        After saving the file, fully quit and restart Claude Desktop (Menu → Quit Claude).
      </Step>
    </Steps>
  </Tab>

  <Tab title="MCP Stdio" icon="terminal">
    First [install Linkly AI CLI](/docs/en/use-cli).

    <Steps>
      <Step title="Open the configuration file">
        In the Claude menu, select **Settings → Developer → Edit Config** to open
        `claude_desktop_config.json`.
      </Step>

      <Step title="Add Linkly AI MCP">
        Add the following configuration to the `mcpServers` object:

        ```json theme={null}
        {
          "mcpServers": {
            "linkly-ai": {
              "command": "linkly",
              "args": ["mcp"]
            }
          }
        }
        ```
      </Step>

      <Step title="Restart Claude">
        After saving the file, fully quit and restart Claude Desktop (Menu → Quit Claude).
      </Step>
    </Steps>
  </Tab>

  <Tab title="Remote Access" icon="globe">
    First complete [remote access configuration](/docs/en/use-in-remote), with the desktop tunnel showing{" "}

    <Badge stroke color="green">
      ● Connected
    </Badge>

    .

    <Steps>
      <Step title="Open Connectors settings">
        In the Claude menu, select **Settings → Connectors**, then click **Add custom
        connector**.
      </Step>

      <Step title="Fill in the endpoint information">
        * Name: `Linkly AI` - URL: `https://mcp.linkly.ai/mcp`
      </Step>

      <Step title="Complete OAuth authorization">
        After adding, you will be automatically redirected to linkly.ai for OAuth authorization. Log in and confirm authorization, then return to Claude to start using it.
      </Step>
    </Steps>
  </Tab>
</Tabs>

<Tip>
  We recommend also installing [Linkly AI Skills](/docs/en/use-skills) to teach the
  AI assistant how to efficiently use Linkly AI's tools, improving search and
  reading efficiency.
</Tip>

## Getting Started

Once configured, ask questions directly in a Claude Cowork conversation, and Claude will automatically call Linkly AI's tools to search and read your local documents. For example:

* "Search my documents for content about project design"
* "Read my requirements document and help me summarize the key points"

## FAQ

<AccordionGroup>
  <Accordion title="Cannot find the Linkly AI tool after restart">
    Please confirm: 1. The configuration file JSON format is correct (you can
    use a JSON validator to check) 2. Claude has been **fully quit** and
    restarted (not just closing the window — quit from the menu) 3. Linkly AI
    Desktop is running with the MCP service active
  </Accordion>

  <Accordion title="Stdio method says it cannot find the linkly command">
    Make sure you have [installed Linkly AI CLI](/docs/en/use-cli) and that the
    `linkly` command is available in your PATH. You can run `linkly --version`
    in the terminal to verify the installation.
  </Accordion>
</AccordionGroup>