Pricing

Inspect and explore the whole public record for free, forever. Pro adds continuous watching and alerts — free for every account until December 31, 2026.

Pro is free for every account until December 31, 2026. Every feature below is live today. Create an account, verify a domain, and start watching. No card required to sign up. No waitlist.

Free — Inspect & history

Works today

$0/month

Free forever. No account required.

“Everything about a public MCP server that is true for one visitor is true for all of them — so all of it is free.”

Inspect any public MCP URL — paste an endpoint; a scanner observes it — nothing runs in your browser
Immediate report for known servers — the public inventory is already under continuous observation
Manual Check now — one fresh observation on demand (cooldown: 60 s public, 30 s private)
Capability inventory & schemas — tools, resources, and prompts with full parameter tables
Full public history & timelines — every observed event since collection began
Structured contract diffs — classified from breaking to informational, with verified evidence
Contract hash — one canonical hash per revision, computed centrally, copyable
Private GUID pages — unlisted endpoints get a private report URL you control
Bookmarkable URLs — every report, diff, and timeline is a real link you can share

Inspect an MCP server Browse the public inventory

Pro — Watch & alert

Free Pro until 12/31/2026

$0/month until 12/31/2026

Then $19/month per account. No card required until then.

“Stop checking. Pro watches your endpoints continuously and tells you the moment something meaningful changes.”

Scheduled background checks — your endpoints observed continuously — no tab open, no reminders
Alerts on meaningful events — confirmed outages, recoveries, contract drift, protocol changes
Email & webhook delivery — signed webhooks and email, where your team already looks
Visual rule builder — “notify me when an optional parameter becomes required”
Register any endpoint — public hosts, internal hostnames, raw IPs — all register instantly, no gatekeeping
Shared public edge scanning — prove a domain by DNS or well-known record to unlock edge probes of its hosts
Customer-hosted scanners — observe endpoints inside your own network with a scoped token
Endpoints saved to your account — registered endpoints, alert rules, and destinations follow your key everywhere (the free browser watchlist stays local to one browser)
Publisher claim & badges — verify ownership, publish official metadata, embed live badges
Private endpoint history — full timelines and diffs for servers that never touch the public inventory
Exports — the raw observation record as one JSON document — GET /api/endpoints/{id}/export

Create an account — free Run your own scanner

Also on the roadmap

Authenticated contract monitoring — arrives with OAuth support
Team access — shared visibility across an organization
CI/CD checks — fail a build when a dependency’s contract drifts

What happens on January 1, 2027

Nothing switches off, and nothing is charged automatically. There is no card on file to charge, so continuing on Pro is a decision you make — never a default that happens to you while you are not looking.

Accounts created before that date get told well in advance and keep grandfathered terms. You will know the number and the date long before either applies to you.

And if you decide not to continue, nothing is deleted. Scheduled watching and alerts stop; inspection, the full public record, every timeline and contract diff stay free forever — exactly as they are today.

Three questions, three answers

Why is public monitoring shared?

An observation of a public MCP server is equally true for everyone, so each public endpoint is observed on one shared schedule and every visitor reads the same accumulated record. That keeps the historical record consistent — and keeps MCP Tools a polite guest: one probe per interval for the whole world, not one per viewer.

Pro scheduling adds your cadence and your alerts on top of the shared record; it never duplicates the public history.

What does “passive inspection” mean?

Scanners speak just enough MCP to take inventory: the initialize handshake (or transport discovery for stateless servers) plus the list calls — tools/list, resources/list, prompts/list.

They never invoke a tool, never read a resource body, and never send arguments. MCP Tools reads the menu; it does not order.

Can I monitor internal endpoints?

Yes — with a customer-hosted scanner. You run a single auditable Python file inside your network; it makes outbound, token-scoped connections to the control plane, claims jobs for your endpoints, observes them locally, and submits results. Your endpoints never need to be reachable from the internet.

See Scanners for the script, the token flow, and the full API surface.

Scanners observe. MCP Tools interprets. You get told what changed.