We score agent-readiness. So we're agent-ready.
Prolectio's whole job is measuring whether an AI agent can actually use a business. It would be strange to fail our own test — so Prolectio ships WebMCP tools on every page. An agent doesn't have to read the screen and guess where the buttons are; the site hands it a labeled set of actions and answers them through Prolectio's own code. The person and the agent share the same page — when the agent runs a tool, you see it happen.
Explore Home · the Prolectio Index · the methodology · pricing · get your free score.
What WebMCP is, plainly
When an AI agent lands on a normal website, it has to read the page like a person — parse the HTML, infer intent, and try to click through. It is slow and it breaks. WebMCP lets a site publish a small menu of typed tools — each with a name, the inputs it needs, and what it returns — that an agent can call directly. Same website; agents can now act through the site's own logic instead of scraping it. It is an emerging web standard (surfaced to the page as navigator.modelContext).
The tools Prolectio exposes
Five tools, registered on every page. Four are read-only; one starts a scan. All run on public data over Prolectio's own public read API — no login, no keys, the same rate limits as the HTTP API.
ask_siteAnswer a question about Prolectio from the site's own content — what it is, how the score works, pricing, the Index, security and crawl conduct.Readget_prolectio_scoreLook up a domain's public Prolectio Score (0–100) and grade band, with the report link.Readget_prolectio_reportGet a domain's full public report — the category breakdown and the specific checks an agent stalled on.Readbrowse_prolectio_indexBrowse the Prolectio Index — the site-wide score distribution and, when published, the named leaders.Readrequest_prolectio_scoreStart a free, no-signup scan for a URL and open its report — the site's core action, run on the visitor's behalf.Action
How to try it
WebMCP is brand-new, so today it works in a WebMCP-capable browser — the ChatGPT desktop app's in-app browser, or Google Chrome with WebMCP enabled. In one of those, open prolectio.com and ask the agent something like:
- “What's the Prolectio Score for stripe.com?”
- “Scan docs.mycompany.com on Prolectio and tell me where an agent would stall.”
- “How does Prolectio calculate the score?”
On any other browser the page works exactly as it always has — the tools stay dormant and nothing changes for a human visitor.
Why this matters
This is the shift Prolectio measures, made concrete. A prospect using an agent can reach a Prolectio Score without ever touching the form — the agent is the on-ramp. And it is the honest version of our own pitch: we don't just tell businesses to be usable by agents, we did it, in public, on the site that grades it. See the methodology for how the score itself is built.
The sharper version: agents trying to buy, not just read
Reading your docs is the first thing an agent tries. Buying from your store is the harder, higher-stakes version of the same problem — and it is where Prolectio's agent-readiness work is headed next. An AI shopping agent (ChatGPT, Perplexity, and similar) can get stuck the same three ways a docs-reading agent does: blocked outright by bot-management it never meant to trip, unable to parse a catalog or complete a cart, or stopped cold at a checkout wall built only for a human browser. Prolectio's rung ladder grades exactly that — from a one-time static conformance read (free, every scan) up through a live agent driving your checkout to the brink on a schedule, taxonomizing precisely where it stalled with a signed evidence transcript, never a bare pass/fail. The Monitor and Monitor+ plans turn that into a recurring check plus a fix, and a sustained clean streak earns a public, machine-readable /verify/{domain} page. We hold ourselves to the same bar the tools above already prove: an agent can act through Prolectio's own logic today; the goal is that any shopping agent can eventually do the same on every store we monitor.