👑

MCP QUEEN

FOUNDING PILOT 10 STACKS MAX

Scout tells you when something better appears. Guard tells you when something you trust has changed.

Give MCP Queen one agent job, your constraints, and the MCP servers you use today. Get a seven-day evidence check before deciding whether continuous monitoring is worth paying for.

This is a deliberately small, concierge-backed pilot. MCP Queen is validating the decision and alert workflow before building accounts, dashboards, policy engines, or automatic enforcement.

Start with a free seven-day Agent Stack Check.

Send the task, current servers, security tolerance, budget, constraints, and preferred platforms. Queen replies with a dated baseline, better-fit candidates when the evidence supports them, explicit unaudited gaps, and the exact changes worth monitoring.

Start my free stack check See the founding price Try Queen in my agent

No card for the free check. Intake happens by email. Do not send credentials, tokens, private data, or proprietary prompts.

One shared evidence loop

Scout: is there a better fit now?

Searches observed tools and public Registry metadata for the job, checks operational evidence and repository mobility, and explains why a candidate may be better—or why the evidence is not strong enough to recommend a switch.

Guard: did the trusted baseline change?

Rechecks public identity, endpoint behavior, observed tools and schemas, repository mapping, and available Trust Receipt evidence. A material delta produces an evidence-linked action recommendation.

The shared MVP is one Stack Brief, not two products.

The brief records the job and constraints, the current shortlist, a Trust Lease preview for each approved server, a dated change log, and one recommendation: continue, warn, review, quarantine, or block in your own execution layer.

What arrives during the free check

Baseline: current public Registry identity, endpoint, observed capabilities, operational evidence, repository mapping, available Trust Receipt dimensions, and every important unaudited gap.
Scout comparison: up to three alternatives that fit the stated task, platform, budget, and tolerance—only when the available evidence supports the comparison.
Guard preview: the material fields to recheck and the action that each kind of change should trigger.
One follow-up: a seven-day delta showing what changed, what did not, and whether ongoing monitoring saved time or changed a decision.

PRICE TEST

Founding Stack Watch — $49/month

One agent stack · up to five public MCP servers · weekly Scout check · material-change Guard alerts · evidence links and explicit unaudited gaps.

Billing starts only if the free brief is useful and you explicitly choose to continue. The founding price is being tested with the first ten stacks; payment received—not an invoice or verbal interest—is the conversion signal.

Claim a free check first I would pay $49/month

What this pilot does not pretend to be

No blanket safety score or security certification.
No inspection of private data, credentials, authenticated tool behavior, or proprietary agent traces.
No automatic blocking, quarantine, or permission changes inside your agent yet.
No promise that a public probe observes every runtime or supply-chain change.

The pilot proves whether dated, evidence-linked recommendations change real MCP selection and approval decisions. Automation expands only after active use and payment validate that job.

Pilot questions

Is this a security certification?

No. The pilot separates operational observations, Trust Receipt dimensions, repository evidence, and unaudited gaps. It does not certify a server or replace source review, least privilege, authorization controls, or private testing.

Does Guard block tools automatically?

Not in the founding pilot. Guard recommends continue, warn, review, quarantine, or block based on dated changes. You keep the execution decision in your own agent or approval layer while MCP Queen validates demand.

What counts as a material change?

A change to public Registry identity, endpoint reachability, observed tools or schemas, access behavior, repository location, operational evidence, or available Trust Receipt evidence. Coverage depends on what the public endpoint exposes; missing evidence stays unaudited.

Do I need to share credentials or agent data?

No. Send only the task profile and public MCP server names or URLs. Do not send credentials, tokens, private prompts, customer data, or proprietary agent traces.