Give your agent access to
1,900+ tool ratings

The Agent Native Registry MCP server lets your AI agent search, compare, and score tools mid-task. No API key needed. One line to install.

Install

One command, any MCP client

Works with Claude Code, Cursor, Windsurf, and any MCP-compatible client. No auth required.

# Claude Code
claude mcp add --transport http agent-native-registry https://agentnativeregistry.com/api/mcp
// .claude/settings.json
{
  "mcpServers": {
    "agent-native-registry": {
      "type": "http",
      "url": "https://agentnativeregistry.com/api/mcp"
    }
  }
}
// Cursor / Windsurf / other MCP clients
{
  "mcpServers": {
    "agent-native-registry": {
      "url": "https://agentnativeregistry.com/api/mcp"
    }
  }
}
Tools available

4 tools your agent can call

Each tool is designed for agents — structured output, clear descriptions, no auth required.

Search Tools

search_tools

Search 1,900+ tools by name, category, or use case. Filter by minimum score. Returns agent-native scores and key metadata.

{ "query": "email API", "min_score": 50, "limit": 5 }

Get Score

get_score

Full score breakdown for any tool. Returns all 5 dimensions (discovery, auth, pricing, tooling, reliability) plus the 33-point checklist.

{ "name": "Stripe" }

Compare Tools

compare_tools

Side-by-side comparison of 2-5 tools. Shows score breakdown, key checks, and a recommendation based on agent-friendliness.

{ "names": ["Resend", "SendGrid", "Postmark"] }

List Categories

list_categories

Browse all tool categories with counts. Use this to discover what's available before searching.

{}
Use cases

When your agent needs a tool

Real scenarios where agents use the registry mid-task.

Choosing between alternatives

"I need to send emails from my app. Should I use Resend, SendGrid, or Postmark?"

Your agent calls compare_tools and gets a structured comparison — scores, auth methods, MCP availability, free tier limits — instead of guessing.

Validating a tool before integrating

"Does this tool have an OpenAPI spec? Can I auth with an API key? Is there a sandbox?"

Your agent calls get_score and gets a pass/fail checklist for all 33 agent-readiness criteria. No more trial and error.

Discovering new tools

"What payment tools work best with AI agents?"

Your agent calls search_tools with a category filter and minimum score. Gets a ranked list of the most agent-native options.

Also available

REST API

Prefer REST over MCP? The same data is available via our JSON API. No auth needed.

# Search tools
curl "https://agentnativeregistry.com/api/tools?category=payments&has_mcp=true"

# Score a tool on-demand
curl -X POST https://agentnativeregistry.com/api/score \
  -H "Content-Type: application/json" \
  -d '{"toolName": "Stripe", "toolUrl": "https://stripe.com"}'

Full API reference: API Docs

Is your tool in the registry?

1,900+ tools scored. Check your Agent Native Score and see how agents experience your tool.

Check your score