Model Context Protocol moderate risk

Kagimcp

The Kagi MCP server provides access to Kagi Search and Summarizer APIs, enabling language models to retrieve information and summarize content.

Connections & Capabilities

Connects To

GitHub

Capabilities

read

Quickstart

Install

npx -y @smithery/cli install kagimcp --client claude

Config

{
  "mcpServers": {
    "kagi": {
      "command": "uv",
      "args": [
        "--directory",
        "/ABSOLUTE/PATH/TO/PARENT/FOLDER/kagimcp",
        "run",
        "kagimcp"
      ],
      "env": {
        "KAGI_API_KEY": "YOUR_API_KEY_HERE",
        "KAGI_SUMMARIZER_ENGINE": "YOUR_ENGINE_CHOICE_HERE" // Defaults to "cecil" engine if env var not present
      }
    }
  }
}

Exposed MCP Tools (2)

safe

search

Retrieves search results from Kagi's search index based on a given query.

Read-only operation, no side effects.

safe

summarize

Summarizes the content of a given URL or text using Kagi's summarization API.

Read-only operation, no direct system modifications.

Safety Assessment

The Kagi MCP server is relatively safe for read-only operations like search and summarization. However, the risk of API key compromise and the potential for biased summarization outputs should be carefully considered. Limiting summarization engine choices and monitoring API usage can mitigate some risks.

Requires an API key for authentication.
Primarily uses read operations for search and summarization.
Does not directly execute code or modify system settings.
Can be configured to limit summarization engine choices.

API key compromise could lead to unauthorized usage.
Summarization of untrusted content could expose the model to biases or misinformation.
Improperly configured summarization engines could lead to unexpected outputs.
Potential for information leakage if queries contain sensitive data.