Do not disclose a vulnerability, exploit, credential, private path, or result body in a public issue.

Use GitHub to privately report a security vulnerability. Include the affected version, impact, reproduction steps, and a minimal proof of concept. Remove unrelated credentials, private paths, and result contents before submission.

Security reports can include:

• bypassing Tool authorization, discovery filtering, or call policy;
• binding a call to the wrong upstream Server or Tool;
• causing one Slim Guard call to execute upstream more than once;
• reading another runtime generation's result snapshot;
• recovering a result that differs from the stored upstream result;
• exposing credentials, arguments, result bodies, or raw capability references through logs or errors;
• bypassing SSRF or parameter restrictions in a configured transport;
• executing code through crafted configuration, Tool metadata, arguments, or results.

• vulnerabilities that exist only in an upstream MCP Server;
• social engineering;
• attacks that require physical access to the Slim Guard host;
• benchmark differences that do not cross a security or data boundary.

When a report is confirmed, the maintainer will coordinate remediation and public disclosure with the reporter. Response and release timing depend on the impact and available evidence.

With the reporter's consent, confirmed contributions can be recognized in the release notes or CHANGELOG.md.