webmcp-tool

Implementation

We fix what the check finds.

Most of what the report flags has copy-paste code attached, and you should take it. The work we do is the part that does not: server rendering an application that was never built for it, designing a tool set worth registering, and getting a discovery document past your own robots.txt.

How it runs

  1. Assessment. The automated check, plus the things it cannot see: authenticated views, the journeys that matter, what your logs say about which agents already arrive.
  2. A written plan. Ordered by points on the table against effort, with what we would not do and why. You can execute it yourself from here, and some clients do.
  3. Implementation. Fitness first — rendering, semantics, structured data — then adoption. In that order every time, because tools on a page an agent reads as blank are tools nobody finds.
  4. Verification. Re-check, plus a real agent run against the journeys we agreed. Not a screenshot of a passing scan.

Scopes

Starter

from €2,500

A site that scores badly on fitness and has adopted nothing.

  • Full agent readiness assessment, beyond the automated check
  • Server rendering for the public pages that need it
  • Form semantics: labels, names, types, autocomplete tokens, error states
  • robots.txt rewritten deliberately, with AI agents named
  • Structured data for the page types that carry your money
  • Five read-only WebMCP tools, with the origin trial configured
  • One month of re-checks after launch

Business

from €6,500

A product with real journeys — a shop, a console, a booking flow.

  • Everything in Starter
  • Around fifteen tools, including conditional tools that appear on login
  • Extraction of business logic out of the view layer where handlers need it
  • A discovery document, verified reachable against your own robots.txt
  • llms.txt generated from your content rather than hand-maintained
  • Agent journey testing across your three most valuable paths
  • Three months of monitoring and re-checks

Enterprise

from €12,000

Regulated, or large enough that a security review is a scheduled event.

  • Everything in Business
  • Read-only tool design for regulated processes, with the threat model written up
  • Security review documentation you can hand to compliance
  • An MCP server where headless callers genuinely need one
  • Governance: who owns the tool catalogue, and how it changes
  • Six months of monitoring, with a named contact

Prices are starting points for a defined scope, not a rate card. A quote follows the assessment.

What we will tell you for free

If your score is low for reasons that are entirely fitness — rendering, semantics, structured data — say so out loud and fix it yourself. It is well-documented work, our guides cover all of it, and paying an agency for it is paying for scheduling. The case for hiring anyone starts when tool design, an unpicking of logic from the view layer, or a compliance conversation is involved.

Where to start

Run the check first. It is free, it takes twenty seconds, and it makes the first conversation a specific one.

One request, about 20 seconds. No sign-up, no email. The result gets a shareable address.

Already know what you need? Talk to us →

For internal tools behind a login or a VPN, the checker cannot reach them by design — that assessment is a conversation and a look at the code. See internal tools.