Security fixes are applied to the latest published version of @brandsystem/mcp and to the current main branch. Older releases may not receive patches; users should upgrade to the latest npm release.

Please report suspected vulnerabilities privately. Do not include technical details, proofs of concept, secrets, or customer data in a public issue.

Preferred method:

1. Open the repository's Security page.
2. Select Advisories.
3. Choose Report a vulnerability.

GitHub Private Vulnerability Reporting keeps the report and subsequent discussion visible only to the reporter and repository security maintainers.

If GitHub reporting is unavailable, email security@brandcode.studio.

Include, when available:

• the affected package version, commit, tool, or endpoint;
• a clear description of the behavior and potential impact;
• minimal steps to reproduce;
• a safely redacted proof of concept; and
• any suggested mitigation.

We aim to acknowledge new reports within two business days. After reproducing and assessing a report, we will share its status and coordinate remediation and disclosure with the reporter. Fix and disclosure timing depends on severity, complexity, and downstream impact.

This policy covers security issues in:

• the published @brandsystem/mcp package;
• source code and hosted MCP behavior implemented in this repository; and
• this repository's build and package-release workflow.

General product support, feature requests, and non-sensitive bugs should use the public issue tracker. Vulnerabilities in third-party products should be reported to their maintainers unless the issue is caused by this project's integration with that product.

Please act in good faith:

• test only accounts, brands, data, and systems you are authorized to use;
• avoid privacy violations, service disruption, data destruction, and unnecessary access or persistence;
• collect only the evidence needed to demonstrate the issue;
• remove secrets and personal or customer information from reports; and
• allow a reasonable opportunity to investigate and remediate before public disclosure.

We will work in good faith with researchers who follow this policy and will credit valid reports when requested and appropriate.

Behavioral scanners (Socket and similar) flag this package for capabilities that are intentional product functions, not vulnerabilities:

Independent facts a reviewer can verify: zero npm advisories on install, zero open Dependabot alerts, registry signatures + SLSA provenance on every release, no install lifecycle scripts in the production dependency tree, and an allowlist-tested tarball (test/package-contents.test.ts) that excludes hosted code, tests, specs, and source maps.