What is an agent-readiness score?

A 0-100 score that measures how well AI agents can discover, understand, use, and pay a product. It covers four layers: Discovery, Accessibility, Usability, and Payments.

How does ora score a website?

ora runs 110+ automated checks against public URLs - no login or self-reporting required. Checks test for MCP server presence, OpenAPI specs, llms.txt, structured data, auth patterns, and more.

What is an MCP server?

Model Context Protocol (MCP) is an open standard for connecting AI agents to external tools and data. An MCP server exposes a product's capabilities as tools that AI agents can discover and invoke.

Ora Directory

About the directory

A short guide to the open formats and services this directory is built on. Each section links to the spec or the live surface for that piece of the stack.

AI Catalog

AI Catalog is a JSON format (application/ai-catalog+json) for describing the resources a product exposes to agents: MCP servers, A2A agents, OpenAPI descriptions, llms.txt indexes, agent skills, and docs. Each entry declares its media type and carries the metadata an agent needs before invoking it.Ora publishes its own catalog at /.well-known/ai-catalog.json and generates one for every scanned product at /d/<domain>/ai-catalog.json.

Ora's AI Catalog

Agentic Resource Discovery

Agentic Resource Discovery (ARD) is an open specification for publishing, indexing, and searching agentic resources across registries. It promotes search-first discovery: a client asks a registry one question over plain HTTP instead of loading every tool description into its context window.This directory is a thin UI over Ora's live ARD registry. The same data is public and CORS-open: search at POST /api/ard/search, facet counts at POST /api/ard/explore, and listing at GET /api/ard/agents. The registry descriptor at /api/ard names every endpoint, filter, and media type.

Agent fronts

Every result in the directory is an agent front: the catalog of one product's invocable resources, synthesized by Ora from a real scan. Ora indexes as an honest third party, so a product's front exists even if its owner never published a catalog. The front reflects what agents can actually use today, not what a vendor self-reports.

Example: stripe.com's agent front

Scores and attestations

Every entry carries the product's agent-readiness score, computed by Ora's open methodology across four layers: discovery, accessibility, usability, and payments.Score attestations are served per domain at /api/ard/attestation/<domain>. Signed attestations verify against the public key set at /.well-known/jwks.json.

How scoring works

← Back to the directory