If you discover a security vulnerability in dns-security-mcp, please report it responsibly.

Email: contact@orhanyildirim.us

Subject line: [SECURITY] dns-security-mcp — <brief description>

• Description of the vulnerability
• Steps to reproduce
• Affected tool(s) or component(s)
• Potential impact assessment
• Suggested fix (if any)

1. Do not open a public GitHub issue for security vulnerabilities
2. Email the details to the address above
3. You will receive an acknowledgement within 48 hours
4. We will work with you to understand and validate the issue
5. A fix will be developed and tested privately
6. A new version will be released with the fix
7. A security advisory will be published after the fix is available

The following are in scope for security reports:

• DNS response injection or manipulation in tool outputs
• Information disclosure through error messages or logs
• Denial of service via crafted inputs
• Dependency vulnerabilities in @modelcontextprotocol/sdk, zod, dns-packet
• Unsafe handling of DNS wire-format data
• Cache poisoning through the internal TTL cache
• Rate limiter bypass
• MCP protocol-level vulnerabilities
• Unsafe use of node:dns, node:tls, node:net, node:dgram built-ins

The following are out of scope:

• Vulnerabilities in external DNS servers or resolvers being queried
• Issues in upstream MCP clients (Claude Desktop, etc.)
• Social engineering attacks
• Findings from tools that are expected to surface security issues (e.g., hijack_dangling_cname detecting real takeover vectors is working as intended)

dns-security-mcp follows these security principles:

• Read-only — tools never modify DNS records, zone files, or server configurations
• No credentials stored — API keys are passed via environment variables, never persisted
• Graceful degradation — missing API keys disable features rather than causing failures
• Input validation — all inputs validated via Zod schemas before processing
• Structured errors — errors return structured messages, never raw stack traces
• Rate limiting — all external API calls go through per-provider rate limiters
• No arbitrary execution — tools cannot execute system commands or arbitrary code

We appreciate security researchers who report vulnerabilities responsibly. Contributors will be credited in the security advisory (unless they prefer anonymity).