Terms of Service

This summary is for convenience. The sections below are the agreement.

• Test only servers you own or are allowed to test. Our agents call real tools, and those calls can change real data.
• A run is your instruction. You are responsible for what it does to your server.
• You own your servers, your data and your reports. We use them only to provide the service.
• A report is our agents' judgement. It is not a certification that a server is correct or secure.

These terms are an agreement between you and Neural Bridge AI, LLC ("Neural Bridge", "we", "us") for your use of MCP Tester at mcptester.dev (the "service"). By using the service, with or without an account, you accept them. Our Privacy Policy explains how we handle personal information.

If you use the service for a company or other organization, you accept these terms for it, and confirm you have the authority to do so. "You" then means both you and that organization.

You must be at least 16 to use the service.

You give MCP Tester the address of an MCP server. It connects, reads the tools the server declares, and sends AI agents to exercise them: they call the tools, check the answers against the server's own declarations and against public directory rules, and write a report.

The service is under active development. Features will change, and some may be added or removed.

You can use the core of the service as a guest, without signing in. A guest's work lives in one browser: lose the cookie and the work is gone, and we delete guest work 30 days after its last visit. Signing in with Google keeps it.

With an account, keep your Google login secure: whatever happens under your account is your responsibility. Tell us at security@mcptester.dev if you think someone else has used it.

This is the term that matters most. You may point MCP Tester only at a server that you own, or that you have the owner's permission to test in this way. That permission has to cover automated testing by AI agents, including calls that create, change or delete data.

Before you start a run:

• Remember that our agents call the server's real tools. A tool that writes, sends, deletes or charges may really do so.
• Use an account on the server that holds test data, not real customer data, wherever you can.
• If someone else operates the server, let them know a test is coming.

We may ask you to show that you are allowed to test a server, and stop runs or suspend your access until you do.

Every request MCP Tester sends to a server during a run is one you asked for by starting that run, made with the login you gave us. The agents' actions are your actions, and you are responsible for their effects on the server, its data and its users.

The agents work under rules meant to keep a run contained: they mark the data they create, avoid tools that send messages, invite people, pay or publish, and try to clean up after themselves. Those rules reduce the risk; they don't remove it. Agents can misread a tool and make mistakes, and a run that is stopped or fails can leave test data behind. Check your server after a run.

Your servers' addresses, the data they return, your logins and your reports are your content. You own it. You give us a limited permission to store, copy, process and transmit it, including to the providers listed in our Privacy Policy, only to provide and secure the service for you.

You confirm that you have the rights needed to let us process this content, including any personal information your server returns.

We don't use your content to train AI models, and we don't allow the model providers that run your tests to do so.

If you send us feedback or ideas, we may use them without owing you anything.

When you connect a server, you may authorize MCP Tester through the server's sign-in screen or paste an API key. You confirm you're allowed to share that login with a third-party service. We use it only to connect to that server and to run the tests you start; it is never used for anyone else's runs. You can revoke it at any time by disconnecting on the server's page, or on the server's own side.

Your servers, runs and balance live in a workspace. A workspace has owners and members:

• Owners rename the workspace, invite and remove people, and change roles.
• Every member can connect and delete servers, start runs, see and delete the workspace's finished runs, and leave.

The servers, runs and balance belong to the workspace, not to the member who added them, and stay with it when that member leaves. Owners are responsible for who they invite.

A signed-in member can share a finished run. Its link then shows that run's report to anyone who has the link, with no sign-in. Every run is private until someone shares it.

Sharing is your decision. Before you share, check that the report reveals nothing you or your server's users wouldn't want shared. Stopping sharing makes the link stop working straight away, but we can't recall copies someone already made.

You must not use the service to:

• test, scan or probe a server you aren't allowed to test;
• attack, overload or disrupt any system, including ours;
• reach systems inside private networks, or get around the limits we place on addresses;
• get around rate limits, spending limits or access controls, or use someone else's account or login;
• break the law, infringe others' rights, or process personal information you have no right to process;
• store or distribute malware, or content that is illegal or abusive;
• resell the service, or copy it to build a competing one.

Runs are paid for from a balance in US dollars, which belongs to a workspace. An owner adds funds through Stripe, which sells them to you as our merchant of record: Stripe takes the payment, adds any taxes that apply and sends the receipt.

• Each run's price is shown before you start it, and that is the price you are charged. We may change prices at any time; a change applies only to runs started after it.
• A run is charged when it starts. A run you stop yourself, and a run your server fails, have still used it.
• If a paid run fails before it shows any results because our service or our AI provider broke down, its price goes back to the balance. A run that uses up its budget, ends without its final report or reaches its time limit is charged, whatever the cause.
• We may give out free runs, such as one free run per person while a daily number lasts. A free run that fails is not given again. We may change or end free runs at any time.
• The balance has no cash value and cannot be withdrawn. Except where the law requires otherwise, purchases are final. A refund we choose to give goes back to the balance, not to your card.
• If a payment is refunded or reversed by Stripe or your bank, we remove the same amount from the balance, as far as it has not been spent.

We work to keep the service running, but we don't promise any level of uptime unless we have agreed one with you in writing. Runs depend on your server, on AI model providers and on our own providers, any of which can be slow or unavailable.

You can stop using the service at any time, and delete your account from the account dialog.

We may suspend or end your access if you break these terms, if your use puts others or the service at risk, or if the law requires it. For a serious breach, especially testing a server you aren't allowed to test, we may act immediately. Otherwise we will tell you first and give you a reasonable chance to fix the problem.

Sections that by their nature should survive the end of this agreement do so, including your content, disclaimers, limitation of liability, indemnity and governing law.

The service is provided "as is" and "as available". To the extent the law allows, we disclaim all warranties, whether express or implied, including merchantability, fitness for a particular purpose and non-infringement.

A report is the judgement of AI agents at the time of the run. It can be wrong or incomplete. It is not a security audit, a certification, or a guarantee that a server is correct, safe, or that every problem was found. The checks against the Claude and ChatGPT directories follow those directories' published rules; passing them doesn't mean a directory will accept your server. MCP Tester is not affiliated with Anthropic or OpenAI.

To the extent the law allows:

• Neither party is liable for indirect, incidental, special, consequential or punitive damages, or for lost profits, revenue or data, even if told they were possible.
• Our total liability for all claims about the service is limited to the greater of what you paid us in the 12 months before the claim, or US$100.

These limits don't apply to liability that the law does not allow to be limited. Some jurisdictions don't allow some of these exclusions, so they may not all apply to you.

If someone brings a claim against Neural Bridge because of your content, a run you started against a server you weren't allowed to test, or your breach of these terms, you will defend us and cover the resulting losses and reasonable costs.

These terms are governed by the laws of the State of Delaware, USA, without regard to its conflict-of-laws rules. Any dispute is heard in the state or federal courts located in Delaware, and both parties agree to their jurisdiction.

If you are a consumer, you also keep the protection of the mandatory laws of the country where you live.

We may update these terms. When we do, we change the date at the top of this page. For a material change, we will tell signed-in users by email or in the product at least 30 days before it takes effect. If you keep using the service after that, you accept the new terms; if you don't accept them, stop using the service and delete your account.

These terms and the Privacy Policy are the whole agreement between us about the service. If a court finds part of them unenforceable, the rest still applies. If we don't enforce a right straight away, we haven't given it up. You may not transfer these terms without our consent; we may transfer them as part of a merger, acquisition or sale of assets.

• These terms: legal@mcptester.dev
• Security issues: security@mcptester.dev

Neural Bridge AI, LLC, 169 Madison Avenue, STE 2529, New York, NY 10016, USA.