# INDEX-AI — Case study

An open specification proposal that makes websites verifiably readable by AI agents.

**Status:** Open source project · Open specification and reference implementation · 1.0-rc2 (RFC).

INDEX-AI defines what an AI agent can reliably access, extract, and budget once it reaches a site, through a progressive conformance ladder: Level 1 (AI Manifest — identity, publisher, freshness, policy), Level 2a (Agent Index — a flat list of content nodes with clean-text endpoints and measured content sizes), Level 2b (Agent Graph — typed relationships between nodes), and Level 3 (a typed MCP query interface).

This portfolio itself implements the standard: see /.well-known/index-ai.json and /agent-index.json. The reference validator, @hardmachinelabs/index-ai-validator, is published on npm and documented at /labs/index-ai-validator.