# Is your site agent-ready?

> Scan any site for AI-agent readiness — the 0–5 level, what each category checks, and the API and MCP tool behind poneva.com/agent-ready.

AI agents are starting to browse, read, authenticate, and buy on the web on a
person's behalf. The **agent-readiness scanner** at
[poneva.com/agent-ready](https://poneva.com/agent-ready) checks how ready your
site is for them: paste a URL and get a **0–100 score** and a **0–5 level**, with
a per-check breakdown and a copy-paste fix for anything that fails. It's free,
needs no account, and runs in a few seconds.

## The 0–100 score

The headline number is the share of applicable checks your site passes, shown as
a gauge from 0 to 100. Each of the four scored categories — Discoverability,
Content, Bot Access Control, and API/Auth/MCP/Skill Discovery — also gets its own
0–100 sub-score and a pass fraction (e.g. `1/4`). Commerce checks run but are
shown separately and don't count toward the score. The **level** below the gauge
is a coarser milestone view of the same result.

## The 0–5 ladder

Your level is the highest rung whose checks all pass — the ladder is strictly
sequential, so you climb it one rung at a time.

| Level | Name | What it means |
| ----- | ---- | ------------- |
| 1 | Basic Web Presence | A valid `robots.txt` and a sitemap |
| 2 | Bot-Aware | You declare AI usage preferences with Content Signals |
| 3 | Agent-Readable | You serve a Markdown representation for agents |
| 4 | Agent-Integrated | You expose an MCP Server Card, API catalog, Agent Skills and WebMCP |
| 5 | Agent-Native | You publish auth metadata and an A2A Agent Card |

## What we check

~20 checks across five categories. Each returns pass, fail, or "not applicable",
with the exact HTTP request and response we saw — so nothing is a black box:

- **Discoverability** — `robots.txt`, sitemap, `Link` response headers, and DNS
  for AI Discovery (DNS-AID).
- **Content accessibility** — Markdown for agents (`Accept: text/markdown`).
- **Bot access control** — AI bot rules in `robots.txt`, Content Signals, and
  Web Bot Auth.
- **API, auth & MCP** — API Catalog, OAuth discovery, OAuth Protected Resource,
  `auth.md`, MCP Server Card, A2A Agent Card, Agent Skills, and WebMCP.
- **Commerce** — the agentic-commerce standards (x402, MPP, UCP, ACP, AP2). These
  run for every store but don't yet change the level.

Open the **Customize scan** panel to narrow the run to a site type (content site
or API/application) or an exact set of checks.

## Fixing what fails

Expand any failing check to see its **Goal**, the exact **Issue** we found, a
step-by-step **How to implement**, and **Resources** — the relevant specs (RFCs,
drafts) plus a **Skill**. Click the Skill to open a full implementation guide
(`SKILL.md`) in a popup you can copy and hand to a coding agent. **Copy prompt**
grabs the fix for one check; **Improve the score** grabs one combined prompt for
everything that's failing. **Audit details** shows the exact HTTP request and
response behind the verdict.

The skills are also published at
`https://poneva.com/.well-known/agent-skills/` so agents can discover and install
them directly.

## Use it from your own tools

The scanner is an API and an MCP tool, so an agent can scan and reason over the
result directly.

Call it over HTTP:

```
POST https://poneva.com/api/agent-ready/scan
Content-Type: application/json

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

Add `"format": "agent"` to get the combined fix prompt as plain markdown instead
of JSON.

Or call the `scan_site` tool on the poneva MCP server — see
[Agents & MCP](https://poneva.com/docs/agents-mcp) for how to connect.