# Code4X — Full System Architecture, Rubric & Capabilities (llms-full.txt)

> Comprehensive documentation for LLMs and autonomous agents analyzing Code4X.

## 1. What is Code4X?
Code4X is a technical evaluation and remediation platform for website readiness in the AI agent era. As search transitions from human click-through blue links to direct synthesis by Answer Engines (Perplexity, ChatGPT Search, Google AI Overviews, Claude, Apple Intelligence), websites face a third audience: autonomous AI software agents.

## 2. The 55-Dimension Rubric (4 Layers)
- **Layer 1: Agent Discovery (15 Checks)**
  - L1-01: llms.txt presence (200 OK at root)
  - L1-02: llms.txt well-formedness (llmstxt.org spec compliance)
  - L1-03: llms-full.txt presence & depth
  - L1-04: agents.json & agents.txt presence
  - L1-05: ai-agent.json card
  - L1-06: MCP Server Card at /.well-known/mcp.json
  - L1-07: API Catalog (RFC 9727)
  - L1-08: Content Signals policy in robots.txt
  - L1-09: Granular AI bot policy covering 13+ agents
  - L1-10: Cloudflare / CDN default-block inspection
  - L1-11: DNS-AID (_agent TXT record)
  - L1-12: Web Bot Authentication key check
  - L1-13: HTML auto-discovery link headers
  - L1-14: XML sitemap + Markdown sitemap
  - L1-15: Canonical URLs & Indexability

- **Layer 2: Agent Readability (21 Checks)**
  - L2-01: Content negotiation (Accept: text/markdown)
  - L2-02: Markdown mirrors per page (.md availability)
  - L2-03: Accessibility tree & semantic markup
  - L2-04: Zero-JS content renderability (SSR vs SPA)
  - L2-05: Cumulative Layout Shift (CLS) stability
  - L2-07: Schema.org entity graph density
  - L2-08: Schema drift prevention (JSON-LD matches visible DOM)
  - L2-10: Strict single-H1 heading hierarchy
  - L2-12: BLUF (Bottom Line Up Front) concise answer openers

- **Layer 3: Agent Interaction (13 Checks)**
  - L3-01: WebMCP form annotations
  - L3-02: WebMCP registered tools
  - L3-04: Semantic form input labels & types
  - L3-05: A2A (Agent-to-Agent) protocol support
  - L3-06: Agent Skills Discovery manifest

- **Layer 4: Representation Monitoring (6 Checks)**
  - Longitudinal multi-engine query tracking, citation attribution, and factual accuracy scoring.

## 3. Endpoints & Integrations
- Website: https://code4x.dev
- Scanner: https://code4x.dev/scan
- MCP Server: https://mcp.code4x.com/sse
- Agents Manifest: https://code4x.dev/agents.json
- Sitemap: https://code4x.dev/sitemap.xml
- Audit Results: https://code4x.dev/audit/latest.json
- Contact: https://code4x.dev/contact