Manual MCP Setup

tip

If your tool has a dedicated App Guide (Claude, Claude Code, Codex, etc.), follow that instead — it will be faster.

This guide covers connecting any MCP-compatible AI tool to Gety.

Step 1: Create an MCP integration in Gety

1. Open Gety → Settings → AI Integrations
2. Under More Options, click MCP
3. Enter a name for the tool you want to connect (e.g. "Gemini") and click Create

Gety: enter the name of your AI tool and click Create

Step 2: Copy the config to your AI tool

After creating the integration, click View Config next to it. Gety shows the full MCP configuration.

Gety: click the copy button to copy the JSON config

Option A: Paste the JSON to your AI tool

The fastest way — copy the entire JSON block and paste it directly into your AI tool's MCP configuration file. Most tools accept this format:

{
  "mcpServers": {
    "gety": {
      "type": "http",
      "url": "http://127.0.0.1:31226/mcp",
      "headers": {
        "Authorization": "Bearer sk-gety-api-xxxxx"
      }
    }
  }
}

Save the file and restart your AI tool.

Option B: Enter the URL and key manually

If your AI tool has a settings UI for MCP servers, enter these values:

Field	Value
Type	Streamable HTTP (or http)
URL	http://127.0.0.1:31226/mcp
Authorization header	Bearer sk-gety-api-xxxxx (from your config)

tip

Some AI tools have a chat interface that can configure MCP for you. Try pasting the JSON and asking: "This is my Gety MCP config. Please add it to your MCP settings."

Step 3: Verify the connection

Ask your AI tool:

Use Gety to search for [any topic you have files about] and summarize what you find.

If it returns results from your local files, the setup is complete.

Troubleshooting

Can't connect to Gety?

• Make sure Gety is running — check the system tray icon
• Check that indexing has started (it does not need to finish — partial results work)

AI tool doesn't seem to use Gety?

• Some tools require you to explicitly enable the MCP server in their settings after adding it
• Try mentioning "use Gety" in your prompt

Results seem incomplete?

• Gety may still be indexing. Check progress in Gety → Home (bottom right corner)
• Semantic search requires the embedding model (~100 MB download on first launch)

Data sharing settings

When an AI tool calls Gety, you can control what gets shared.

Go to Gety → Settings → AI Integrations → Data Sharing Control:

Gety: choose how data sharing is handled

• Ask every time — A confirmation popup appears before each request. Click Allow to share the snippets, or Don't allow to block the request.

  

  Gety: review the content before allowing access

• Always allow — Content is sent directly without confirmation, for a faster workflow.

You can change this setting at any time.