Open Source · Agent-Readiness Toolkit

Agent-Readiness Scan

A free, open-source CLI that scores how ready your website is for AI agents — not whether it looks good to a human, but whether an autonomous agent can discover you, trust you, and act on your behalf. One command, a 0–100 score with a letter grade, and a concrete fix list.

$ npx @kirkelabs/agent-readiness-scan https://yoursite.com

Why agent-readiness?

Buyers increasingly arrive through an AI agent, not a browser. The agent reads the web on their behalf, decides whether you’re credible, and — increasingly — tries to transact. The modern web is becoming a customs house: agents need papers to pass, and your site needs to issue and honour them. Most sites are invisible or unactionable to this traffic and don’t know it.

The scan measures the signals that decide the outcome — across eight weighted dimensions — and tells you exactly what to add.

What it scores

Per-bot crawler policy. Which AI crawlers you name, and what you allow each.
Declared use-policy. Content Signals: search / AI-input / AI-training, on the record.
Bot-Auth readiness. Can verified agents be told apart from impostors (Web Bot Auth)?
MCP exposure. Do you publish agent-callable tools — can agents act, not just read?
Agentic-commerce. Can an AI agent transact on your behalf (ACP / UCP / x402)?
Product / Offer. Are your offers structured for agent-actionable checkout?
Brand identity. Registry-grade corroboration (Wikidata, Crunchbase, LEI…).
Source provenance. Dates, security.txt, terms, privacy — the trust paperwork.

We run it on ourselves

We built it, then scanned kirkelabs.com — and scored a humbling 29/100 (F). So we shipped the fixes in the open: per-bot Content Signals, a Web Bot Auth key, a security.txt, structured-data dates, terms & privacy, an agentic-commerce manifest, and a live Open Agent Access MCP server — taking the score to 73 and climbing. The tool is only worth shipping if it survives being pointed at us first.

Run it on your own site and see what an agent sees.

Part of the agent-readiness toolkit

Two open-source scanners, two questions every site should be able to answer:

agent-readiness-scan — can agents act on you? Discovery, trust boundaries, tools, and commerce. GitHub
ai-legibility-scan — can LLMs read you? How legible your pages are to AI agents and language models. GitHub

Open-source (run, read, contribute): github.com/KirkeLabs/agent-readiness-scan · npm