Skip to main content

Lumar Analytics Plugin

The open-source lumar-analytics plugin packages the Lumar MCP connection with reusable skills for common Analyze and AI Visibility workflows. It supports:

  • Claude Code
  • Cursor
  • OpenAI Codex

The plugin is optional. Direct MCP connections still expose the same server tools, while the plugin helps an agent choose and sequence those tools consistently.

What the plugin adds​

The plugin:

  • Configures the lumar MCP server at https://mcp.lumar.io/mcp.
  • Shares one skills tree across Claude Code, Cursor, and Codex.
  • Guides account/project resolution, pagination, asynchronous operations, confirmation of write actions, and product-specific caveats.
  • Provides focused workflows instead of requiring users to know individual MCP tool names.

Analyze workflows​

The plugin covers crawl-health audits, report deep dives, URL investigation and stored HTML, exports, task management, crawl runs, segments, Single Page Requester, custom metrics, Jira links, and project administration.

AI Visibility workflows​

The plugin covers AI Visibility audits, competitor benchmarking, project/topic setup, prompt investigation, score trends, page evaluation, brand curation, Google Search Console setup, and provider management.

Install in Claude Code​

  1. Add the marketplace:

    /plugin marketplace add deepcrawl/lumar-plugins

  2. Run /plugin, open Marketplaces, choose lumar-plugins, and install lumar-analytics.

  3. Run /mcp and complete the browser-based Lumar authentication flow.

Install in Cursor​

  1. Open Cursor's plugin manager.
  2. Add deepcrawl/lumar-plugins as a marketplace source.
  3. Install lumar-analytics.
  4. Open the MCP settings, find the provisioned lumar server, and complete authentication.

You can also point Cursor at a local clone of the repository when testing a plugin change or a different MCP endpoint.

Install in Codex​

  1. Add the marketplace from your shell:

    codex plugin marketplace add deepcrawl/lumar-plugins

  2. Run /plugins in Codex and install lumar-analytics.

  3. Complete the Lumar OAuth flow when Codex prompts during installation.

Use a skill​

Skills trigger from normal requests. For example:

Audit the latest crawl for <project> and rank the issues that grew most.
Investigate <url>, including the stored rendered HTML and its top accessibility issues.
Compare our AI Visibility with tracked competitors over the last 30 days.
Create a product-page custom metric, test it on these URLs, and prepare it for this project.

The agent may ask for an account, project, or confirmation before continuing. This is expected when several resources match or the workflow performs a write action.

Use a local plugin checkout​

Clone the repository:

git clone https://github.com/deepcrawl/lumar-plugins.git

Then add the checkout as a local marketplace using the host's marketplace command or UI. A local checkout is useful for:

  • Testing pre-release plugin changes.
  • Pointing at staging or a self-hosted MCP server.
  • Selecting toolsets through a scoped URL or X-MCP-Toolsets header.
  • Supplying a headless credential through a local, uncommitted configuration.

Never commit user keys or service-account keys to the plugin repository.

For configuration details, see Authentication and toolsets. For the complete plugin source and current skill inventory, see deepcrawl/lumar-plugins.