MCP 2.0: 30 Tools for Claude Desktop, Cursor, and Windsurf

Today we shipped AI Sightline MCP 2.0. The public npm package now exposes 30 tools, up from 10. If you use Claude Desktop, Cursor, Windsurf, or any other MCP client, your AI assistant can now read, reason over, and act on your entire AI visibility footprint, not just a slice of it.

The headline number is simple. We tripled the tool surface in a single release.

Why MCP is the right interface for AI visibility

A dashboard shows you a number. An MCP server lets your AI agent reason about it.

When your visibility data lives inside Claude Desktop, the workflow stops being "log in, filter, export, think." It becomes "ask Claude what to fix this week." The dashboard is still there when you need it. But the day-to-day work, deciding what to prioritize, drafting briefs, spotting sentiment drift, happens inside the assistant you are already using.

We built the MCP server because this is where technical marketers and SEO operators actually work. Not in another browser tab.

What landed in 2.0

The ten tools that shipped in 1.x covered the basics: read your score, list your prompts, pull your mentions. Useful, but narrow.

2.0 closes the gap between the internal product and what the public package can do. The new tools break down into seven capability groups.

Multi-brand management. list_brands, create_brand, archive_brand, reactivate_brand. If you run more than one brand, or run an agency, you can now manage the full brand roster without leaving your client.

Quota visibility. get_org_quota tells your assistant how much of your plan you have left. No more guessing why a tool call returned a quota error.

Historical trends. get_trends pulls visibility over time so Claude can tell you what changed this week versus last, with numbers.

Topic authority. get_topic_authority surfaces which topics you actually own across AI platforms, and which ones you do not.

Brand sentiment. get_brand_sentiment exposes our BSI pipeline so your assistant can read not just whether you were cited, but how you were framed.

Recommendations engine. The full recommendations pipeline is now queryable. Your assistant can pull your top-ranked actions, explain the reasoning, and track which ones you have applied.

Content and schema. Domain reporting, content audit, content generation, schema suggestions, and sentiment trends. The full content intelligence stack, previously dashboard-only, now available from the command line of your AI client.

Three workflows worth trying tonight

Paste any of these into Claude Desktop or Cursor after you install the MCP:

1. "Pull my top 5 GEO recommendations, then draft a 300-word content brief for each one."This chains the recommendations engine into the content generator. You get briefs your team can pick up tomorrow morning.

2. "Compare this week's visibility trend to last week across all my brands. Flag anything that dropped more than 10 points."This uses list_brands, get_trends, and a bit of reasoning. Perfect Monday-morning check-in.

3. "Show me my sentiment trend on Perplexity for the last 30 days and explain what changed."get_brand_sentiment plus sentiment trends. Surfaces narrative drift before your CMO asks about it.

Install

The package is on npm as @aisightline/mcp-server (v1.10.0). In your MCP client's config, point it at the package and pass your API key:

{
"mcpServers": {
"aisightline": {
"command": "npx",
"args": ["-y", "@aisightline/mcp-server"],
"env": {
"AISIGHTLINE_API_KEY": "your_key_here"
}
}
}
}


Grab your API key from the Developer → API Keys section of the portal sidebar (aisightline.com/portal/api-keys).

Full install docs including config snippets for Claude Desktop, Claude Code, Cursor, and Windsurf, plus every tool the server exposes are live at aisightline.com/portal/api-docs#mcp.

What is next

MCP is how we think AI visibility tooling should feel by default: not a dashboard you visit, but a capability your AI already has. 2.0 is the biggest single jump we have shipped on that path. The next release focuses on write-path tools so your assistant can not only read your visibility data but also act on it (adding prompts, applying recommendations, running experiments).

If you are already a customer, update the package and you are done. If you are not, the free tier includes read-only MCP access, which is enough to try most of what we shipped today.

Start monitoring your AI visibility, free, no credit card required. See your score in under a minute at aisightline.com.