Terms of use

The service

This server, run by Maël Belliard, checks AI agent skills and MCP servers for security risks before you install them: it scans what you submit with skillspector, and, if you ask, has an AI model review it. Using it means accepting these terms. The software is open source, under the MIT licence; these terms are about this server.

Your account

• An account is for one person. Give an email address you can receive mail at.
• Keep your password, API tokens and keys to yourself: you’re responsible for what’s done with them. Revoke a token or change your password as soon as you think it leaked.
• You can delete your account at any time, from your Account page.

Acceptable use

Submitting suspicious skills to have them checked is what the service is for. Using it against the service, its providers or anyone else is not. In particular, don’t:

• submit content you have no right to access or have analysed;
• try to get around the quotas, rate limits or bot protection, for example with several accounts, or overload the server;
• try to reach other users’ data, the server’s secrets or systems, or use scans to reach other systems than the code hosts they’re meant for;
• use the service for anything unlawful, or put personal data or secrets of others in what you submit without reason.

Found a vulnerability? Please report it privately to contact@maelbelliard.fr rather than exploit it.

Quotas

The operator can limit how many scans each account starts a day and runs at once, and how often requests can be made. Your Account page shows your current limits. They can change, and scans can be paused for everyone, for example during an incident.

Results are advice, not a guarantee

Scans are automated, by static analysis and, when you turn it on, an AI model. They can miss a real threat and flag harmless code. A “safe” verdict means nothing alarming was found, not that the skill is safe. You remain responsible for what you install and run: read a skill’s code, and weigh a result with your own judgement, before trusting it.

What you submit

• You keep all your rights over what you submit. You allow the operator to fetch, store and analyse it, and to show the results to you and to whoever you share them with, only to provide the service.
• With AI review on, the skill’s content is sent to the AI provider you chose, under its own terms.
• A result you share, or put on a badge, is public: make sure you may publish it, especially for a private repository.

Availability

The service is provided as it is, without any guarantee of availability. A scan can fail or time out. The operator can change it, suspend it or end it, and will try to give notice of an end, so you can download your reports.

Liability

As far as the law allows, the operator isn’t liable for indirect damage, or for losses from relying on a result, a scan that failed, or the service being unavailable. Nothing in these terms limits liability that the law doesn’t allow to be limited, nor the rights consumers have under the law.

Suspension

The operator can suspend or delete an account that breaks these terms, after telling its owner why where possible, or at once for a serious breach such as an attack on the service.

Changes and law

These terms change when the service does; the date at the top says when they last did, and using the service afterwards means accepting them. They’re governed by French law and the French courts. Questions: contact@maelbelliard.fr.