Google PageRank for AI agents. 25,000+ tools indexed.

How it works

AgentRank publishes an MCP server (agentrank-mcp-server on npm) that gives any MCP-capable agent direct access to the live ranked index. Platforms that don't support MCP natively get an instruction file that teaches the AI to query the REST API instead. Either way, your agent stops guessing and starts looking things up.

Cursor

MCP native

Cursor supports MCP servers natively. Install once and Cursor will use AgentRank automatically when you ask for tool recommendations.

Option 1 — MCP server (recommended)

Add to ~/.cursor/mcp.json for global use, or .cursor/mcp.json in your project for project-scoped access.

mkdir -p .cursor && curl -o .cursor/mcp.json https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/cursor/mcp.json
Show config
{
  "mcpServers": {
    "agentrank": {
      "command": "npx",
      "args": ["-y", "agentrank-mcp-server"]
    }
  }
}

Option 2 — Rules file

If you prefer cursor rules over MCP, copy .cursorrules to your project root or ~/.cursorrules for global use.

curl -o .cursorrules https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/cursor/.cursorrules

For the new .cursor/rules/ format (Cursor 0.43+):

mkdir -p .cursor/rules && curl -o .cursor/rules/agentrank.mdc https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/cursor/rules/agentrank.mdc

VS Code + Copilot

MCP native

VS Code 1.99+ supports MCP servers in agent mode. Add AgentRank to your workspace and Copilot will query it automatically.

Option 1 — MCP server (recommended)

Add to .vscode/mcp.json in your project. VS Code picks this up automatically in agent mode.

mkdir -p .vscode && curl -o .vscode/mcp.json https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/vscode/mcp.json
Show config
{
  "inputs": [],
  "servers": {
    "agentrank": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "agentrank-mcp-server"]
    }
  }
}

Option 2 — Copilot instructions

Copy copilot-instructions.md to .github/ — Copilot reads this automatically for workspace-level context.

mkdir -p .github && curl -o .github/copilot-instructions.md https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/vscode/copilot-instructions.md

Cline

MCP native

Cline supports MCP servers. Add AgentRank via the Cline MCP settings in VS Code, or drop a rules file for instruction-based access.

Option 1 — MCP server (recommended)

Open Cline in VS Code, go to MCP Servers, click Configure MCP Servers, and add the following to cline_mcp_settings.json:

{
  "mcpServers": {
    "agentrank": {
      "command": "npx",
      "args": ["-y", "agentrank-mcp-server"],
      "disabled": false,
      "autoApprove": ["search", "lookup", "get_badge_url"]
    }
  }
}

Option 2 — Rules file

Copy .clinerules to your project root or ~/.clinerules for global use.

curl -o .clinerules https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/cline/.clinerules

Claude Code

MCP native

Claude Code has native MCP support. One command installs the server globally.

MCP server

claude mcp add agentrank -- npx -y agentrank-mcp-server

Or install the skill (includes both MCP and instructions):

npx skills add superlowburn/agentrank --skill skill/agentrank

Windsurf

Rules file

Windsurf uses .windsurfrules for AI instructions. Copy to your project root or global rules location.

Rules file

curl -o .windsurfrules https://raw.githubusercontent.com/superlowburn/agentrank/main/skill/agentrank/windsurf/.windsurfrules

For global use, add to ~/.codeium/windsurf/memories/global_rules.md.

Claude Desktop

MCP native

Add to your Claude Desktop config to use AgentRank in any Claude Desktop conversation.

MCP server

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "agentrank": {
      "command": "npx",
      "args": ["-y", "agentrank-mcp-server"]
    }
  }
}

Restart Claude Desktop after saving.

Available tools

Once installed, your agent has access to three tools:

search

Search 25,000+ MCP servers, agent tools, and skills by keyword. Returns scored results ranked by real GitHub signals.

e.g. "browser automation", "database", "slack"

lookup

Check if a specific GitHub repository is in the index. Returns score, rank, and signal breakdown.

e.g. github.com/microsoft/playwright-mcp

get_badge_url

Get an embeddable SVG badge showing a tool's AgentRank score. Paste into your README.

e.g. slug: "glama-playwright-mcp-server"