# agent-readiness

## What is Agent Readiness?

**Agent Readiness** is the degree to which a website supports open standards for interacting with AI agents. The term covers two aspects: visibility to AI systems ([GEO](/glossary/geo)) and the ability of agents to interact with the site's functionality.

A site with high agent readiness can:

1. **Be discovered** — via [robots.txt](/glossary/robots-txt), [sitemap](/glossary/sitemap), [llms.txt](/glossary/llms-txt), [ai-agent.json](/glossary/ai-agent-json), Link headers
2. **Be read** — in structured format: [Schema.org JSON-LD](/glossary/schema-org), [Markdown for Agents](/glossary/markdown-for-agents)
3. **Declare its policy** — [AI bot rules](/glossary/ai-bot-rules), [Content Signals](/glossary/content-signals)
4. **Provide API access** — OAuth, MCP, Agent Skills, A2A
5. **Accept payment from an agent** — x402, MPP, UCP, ACP

The analogy: in 2015, a site had to be **mobile-ready**; in 2026, it needs to be **agent-ready**.

## Why does a site need Agent Readiness?

Three parallel trends make agent readiness a necessity today.

**GEO: a new search channel.** ChatGPT, Perplexity, YandexGPT, and Gemini are replacing search for a portion of users. In B2B and technical topics, this already accounts for 15–30% of queries. A site without AI optimization loses visibility in this channel.

**Agentic workflows: agents as executors.** AI agents perform tasks on behalf of users — they place orders, gather data, and work with APIs. A site without MCP or Agent Skills is inaccessible to them.

**AI-driven development.** Cursor, GitHub Copilot, and Claude Code consume APIs and documentation through AI interfaces. An MCP server and annotated API documentation lower the integration barrier.

## How to implement Agent Readiness?

Agent readiness is built iteratively — from basic to advanced.

**Step 1 — Basic discoverability (1–3 days):**
- Add AI-bot rules, a Sitemap directive, and a Content-Signal to `robots.txt`
- Publish `llms.txt` with navigation
- Publish `/.well-known/ai-agent.json`

**Step 2 — Structured content (1–2 weeks):**
- Add Schema.org JSON-LD to key pages
- Implement Markdown for Agents (content negotiation)

**Step 3 — Agentic API access (2–3 weeks):**
- Publish an MCP Server Card
- Implement OAuth discovery
- Add an Agent Skills index

**Step 4 — Full optimization (1–2 months):**
- Restructure the API around agentic patterns
- Add an A2A Agent Card
- Implement RateLimit headers

## How do we check Agent Readiness?

The scanner measures agent readiness through 23 checks across 5 categories with weighted scoring:

| Category | Weight | Checks |
|---|---|---|
| Discoverability | 20% | 5 |
| Content Accessibility | 25% | 2 |
| Bot Access Control | 15% | 3 |
| API / Auth / MCP | 30% | 8 |
| Commerce | 10% | 4 |

The final score (0–100) maps to levels 1–5. A score above 80 corresponds to levels 4–5 (agent-ready).

[← All glossary terms](/en/glossary)