# LEGIT AGENTIC

LEGIT AGENTIC verifies whether AI systems can discover, understand, and trust a business website.

## Run the verifier

Use the interactive audit at [legit-agentic.vercel.app](https://legit-agentic.vercel.app/). The audit is read-only, samples up to five public pages, and does not log in or submit forms.

## What the verifier checks

- Access: successful HTML responses, robots rules, AI crawler access, and content negotiation.
- Discovery: sitemap, titles, descriptions, Open Graph metadata, canonical URLs, and llms.txt.
- Extractability: substantive raw HTML and maintained Markdown versions.
- Identity: structured data, organization identity, and authoritative sameAs connections.
- Trust: authorship, freshness, evidence density, image alternatives, and security contact.
- Action: semantic navigation, descriptive links, and essential interactions that do not depend on widget JavaScript.

## The agent-readable standard

1. Publish an llms.txt that maps retrieval tools to canonical pages.
2. Inspect origin or CDN server logs because analytics may not record bot traffic.
3. Serve an equivalent Markdown version of every important public page.
4. Honor `Accept: text/markdown` on canonical URLs.
5. Keep essential content and actions usable without JavaScript-only widgets.

## Important limitation

The discoverability score verifies public technical prerequisites. It does not measure or predict live mentions, citations, rankings, or recommendations inside ChatGPT, Claude, Gemini, Perplexity, or another answer engine. Those outcomes require repeated observational testing.

## Source and methodology

The verifier is open source. See the [LEGIT AGENTIC repository](https://github.com/Vicorico17/LEGIT-AGENTIC) for the methodology, CLI, tests, and scoring implementation.

Agents and software can inspect the [OpenAPI contract](https://legit-agentic.vercel.app/openapi.json) for the public audit endpoint.