# AgentEye — Full reference (llms-full.txt)

> This is the extended, machine-readable description of AgentEye (https://agentready-eosin.vercel.app). It contains the product description, the complete scoring model, the scan API reference and pricing so agents can answer questions about AgentEye without crawling the HTML pages.

## What AgentEye is

AgentEye is a web tool for the agentic web. It offers two layers:

1. **Free scan (readability):** Can agents discover and read your site? Produces a 0-100 Agent Readiness Score.
2. **Agent Usability Audit (paid):** Can autonomous AI agents actually use your product? Agents attempt real user tasks with multiple runs, screen recordings, failure traces and a prioritized fix roadmap. Positioning: UserTesting for AI agents — not classic QA automation.

There are two steps in the product journey:

1. **Step 1 — Read (free scan):** Can agents read your site? The free scan checks discovery files, crawlability, structure, content clarity, docs and trust signals, and produces a 0-100 Agent Readiness Score with a full pass/fail diagnosis.
2. **Step 2 — Act (Agent Usability Audit, paid):** Can agents use your site? Autonomous agents attempt real tasks (sign up, book a demo, complete checkout) with repeated attempts, recordings and failure traces. Pricing: Quick Test from $99, Full Agent Audit $349, Continuous testing custom/monthly.

## How the free scan works

- You enter a URL on https://agentready-eosin.vercel.app/. No signup required.
- The scanner fetches up to ~16 public URLs: the homepage, /llms.txt, /llms-full.txt, /.well-known/llms.txt, /robots.txt, /sitemap.xml and common pages such as /about, /docs, /developers, /api, /pricing, /faq, /contact, /legal, /terms and /privacy.
- No login, no deep crawling, no JavaScript execution — the scan sees your site the way a text-reading agent does.
- The result is a permanent, shareable report at https://agentready-eosin.vercel.app/report/[domain] with a score, category breakdown and share image.
- Optionally you can request a free full fix report by email with step-by-step instructions for every failed check.

## Scoring model (7 categories, 100 points)

### Agent Discovery (20 points)
- llms.txt exists (10)
- llms-full.txt or extended markdown resource exists (5)
- .well-known/llms.txt exists (3)
- llms.txt contains useful links/headings (2)

### Crawlability & Robots (15 points)
- robots.txt exists (4)
- robots.txt does not block all crawlers (5)
- homepage returns status 200 (3)
- important pages accessible without login (3)

### Sitemap & Structure (10 points)
- sitemap.xml exists (5)
- sitemap contains multiple useful URLs (3)
- canonical tags or clear meta structure (2)

### Content Clarity (20 points)
- homepage has clear title and meta description (4)
- homepage contains a clear product/company description (5)
- about page exists (3)
- FAQ or explanatory content exists (3)
- content is not only hidden inside JavaScript (5)

### Docs / API Readiness (15 points) — for SaaS and generic sites
- docs page exists (4)
- developer or API page exists (4)
- OpenAPI/spec/reference links found (4)
- markdown or text-based docs available (3)

For local businesses this category is replaced by Local Business Signals (structured data, address, opening hours, phone, menu/services, booking).

### Trust & Canonical Facts (10 points)
- company/contact/legal pages exist (3)
- privacy/terms pages exist (3)
- official links/socials are clear (2)
- canonical facts are easy to extract (2)

### Risk, Claims & Compliance (10 points)
- risk/disclaimer language exists (3)
- no excessive unsupported claims detected (3)
- pricing/limitations are clear (2)
- deprecated/outdated info is marked clearly (2)

Score bands: 0-30 Not Agent Ready, 31-60 Partially Agent Ready, 61-80 Mostly Agent Ready, 81-100 Agent Ready.

## Scan API reference

Endpoint: POST https://agentready-eosin.vercel.app/api/scan

Request body (JSON):

    {
      "url": "https://example.com"
    }

Response (JSON):

    {
      "reportUrl": "/report/example.com"
    }

The report page at https://agentready-eosin.vercel.app/report/[domain] renders the score, category breakdown and full diagnosis. A dynamic share image is available at https://agentready-eosin.vercel.app/api/report-card/[domain] and an embeddable SVG badge at https://agentready-eosin.vercel.app/badge/[domain]. Full documentation: https://agentready-eosin.vercel.app/docs

## The Agent Test (paid)

### Quick Test — $99 (1 tasks)

Best for a fast reality check.

- 1 core user task
- 3 agent attempts
- 1 persona
- Screen recording
- Failure summary
- Top 3 fixes

### Full Agent Audit — $349 (5 tasks)

Best for a complete agent usability review.

- 5 core user tasks
- 15+ agent attempts
- Multiple personas
- Screen recordings
- Screenshots
- Failure traces
- Prioritized fix roadmap
- Re-test after fixes

### Continuous Agent Testing — Custom / monthly

Best for teams that want agent tests after every deploy.

- Recurring task runs
- Deploy monitoring
- Regression detection
- Alerts when agents fail critical flows
- Monthly agent usability report

Higher tiers: Agent Readiness Audit (from $2,500) for a deep manual review, and an Implementation Sprint where we implement the fixes together with your team. Request via https://agentready-eosin.vercel.app/agent-test or https://agentready-eosin.vercel.app/contact.

## Canonical facts

- Product name: AgentEye
- URL: https://agentready-eosin.vercel.app
- Category: Business / web tooling (SaaS)
- Free tier: yes (the scan is free, no signup)
- Paid offering: Agent Test $99–$349, audits from $2,500
- Contact: contact@agenteye.io
- Legal: https://agentready-eosin.vercel.app/impressum, https://agentready-eosin.vercel.app/privacy

## Disclaimer

The Agent Readiness Score is a heuristic diagnosis based on publicly accessible files and pages. It does not guarantee ranking, citation or agent behavior. Scores can change as agent ecosystems evolve.