mcp-from-openapi loads and processes OpenAPI documents — including, by design, documents you may not fully trust. Security reports are taken seriously and handled with priority.

Please do not open a public issue for security reports.

Report vulnerabilities privately via GitHub Security Advisories (preferred), or by email to info@agentfront.dev with the subject [SECURITY] mcp-from-openapi.

Please include:

• A description of the vulnerability and its impact
• A minimal reproduction (spec snippet, code, or PoC)
• The affected version(s)

You can expect an acknowledgment within 3 business days and a triage verdict within 7 business days. Confirmed vulnerabilities are fixed in a patch release with a GitHub advisory (and CVE where applicable) crediting the reporter, unless anonymity is requested. Please give us reasonable time to release a fix before public disclosure.

Security-relevant surfaces of this library include:

• SSRF protection for spec-URL loading and external $ref resolution: DNS resolution with connection pinning, internal/private address blocking (including cloud metadata endpoints), and per-hop redirect re-validation (src/ssrf.ts). See docs/ssrf-prevention.md.
• secureDefaults: true — the recommended one-flag posture when loading untrusted specs (redirects off, external $ref resolution disabled).
• Request building — header/cookie injection guards and RFC 7230/6265 validation in buildHttpRequest.
• Overlay application — prototype-pollution guards on JSONPath matching and merging.

Prior advisories for this package are listed under GitHub advisories and on Snyk.