What is an agent-readiness score?

A 0-100 score that measures how well AI agents can discover, authenticate with, and use a product. It covers five layers: Discovery, Identity, Auth & Access, Integration, and User Experience.

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.

Layer 1 - Discovery

AEO / GEO scoring, from a public URL.

Are AI answer engines finding, citing, and recommending you? Discovery is the layer where ora measures Answer Engine Optimization and Generative Engine Optimization - category share of voice, brand-name search accuracy, citation quality, and training-corpus footprint. Every check is computed from your public web surface alone, with no login and no tracker.

What Discovery measures

Discovery is weighted 20 of 100 in the overall ora score. Every check is computed from a public URL with no authentication, no self-reporting, and no panel data.

CheckPoints

Category share of voiceAEO

Run a category prompt across multiple AI search engines in parallel; rank where the product appears against competitors and report a recommended% score.

Developer resource discoverabilityAEO

Brand-name search returns the product's developer resources (API docs, OpenAPI spec, dev portal). The cleanest single AEO signal.

Brand name discoverabilityAEO

Clean brand-name search (no developer keywords): does the product's domain surface in the top results? The baseline for any implicit AI discovery.

Wikipedia / Wikidata entity presenceGEO

Verified Wikipedia article + Wikidata entity with P856 back-reference to the domain. Wikipedia is the single largest source of citations in AI answers.

Listed in MCP registries

Smithery, mcp.so, Glama, PulseMCP.

NPM/PyPI SDK package

Published SDK or CLI package matching domain or product name.

Agent platform configs

Published agent rules/configs (.claude/, .cursor/, .windsurf/) on GitHub.

Listed on skills.sh

Official skills published on the skills.sh agent skills directory.

Semantic search indexabilityGEO

Heading hierarchy, paragraph density, and content-to-markup ratio for vector embeddings used by retrieval-augmented answers.

Training corpus footprintGEO

Brand mentions across Reddit, YouTube, Wikipedia, third-party reviews - whether the brand has enough open-web presence to land in LLM training data.

Competitive positioning clarityAEO

Differentiators, comparison pages, value-prop language agents would cite when recommending you over alternatives.

Citation quality vs. mentionAEO

When found, is the brand cited as a clickable source vs. a name-drop? Measures RAG extractability.

Fresh-encounter recallAEO

Can a clean session - no connector, no API key - describe the product accurately?

Knowledge cutoff coverageGEO

Ask frontier models about the brand with no retrieval, no tools, no live web - can they describe the product from training memory alone? Direct test of training-corpus residency.

Third-party citation corpusGEO

Sample external content on Reddit, GitHub, Stack Overflow, technical blogs, and tech press; rate the depth and diversity of the corpus future training cuts will absorb.

How ora is different

The enterprise answer-engine analytics category - tools that track brand visibility inside ChatGPT, Claude, Perplexity, and Gemini - has emerged in the last eighteen months. ora covers the same AEO/GEO measurement surface and four additional layers, with a few deliberate differences.

Public URL only

No login. No tracker. No browser-extension panels. No CDN log integration. Every check is reproducible from the open web, which is exactly the surface answer engines see.

Every domain in our index, scored automatically

Enterprise answer-engine analytics tools price by tracked-prompt count and gate the interesting features (full engine coverage, prompt-volume datasets, agent analytics) behind custom plans. ora gives every domain a public AEO/GEO score and a public history.

AEO/GEO is one of five layers, not the whole product

Visibility tells you whether agents find you. ora also scores whether agents can understand you (Identity), authenticate (Auth & Access), integrate (Agent Integration), and complete a task with you (User Experience). Visibility alone is a vanity metric if the auth flow blocks the agent at minute two.

Open methodology

Every check is documented, every weight is in source, the scoring formula is in the repo. No black-box visibility score - you can see why your number is what it is and what changes will move it.

How it relates to Identity

Discovery answers can the answer engine reach you. Identity answers once it does, can it understand you. The two layers are coupled - a great llms.txt with a hostile robots policy still loses Discovery, and great Discovery with no JSON-LD still loses Identity. ora ranks them separately so you can fix the stage that's actually limiting you.

See the full methodology for check-by-check scoring and per-layer weights, or run a scan from the homepage to see your AEO/GEO breakdown.