Skip to content

Security: safesploitOrg/bind9-zone-generator

Security

SECURITY.md

Security

Runtime model

Both conversion directions process pasted text locally. Production modules contain no fetch/XHR/WebSocket, telemetry, third-party assets or browser persistence. Hosting may retain ordinary request logs; extensions and user-created clipboard/download copies are outside the app's retention control.

The HTML declares a meta CSP: local scripts/styles/images, connect-src 'none', blocked objects/forms/base URLs and no-referrer. No inline scripts/handlers or HTML insertion are required. Hosting response headers and browser-native CSP enforcement were not verified here; frame-ancestor policy requires hosting headers and is not claimed.

There is no backend, authentication/session system, database, secret store or DNS/hosts installation authority. The downloaded hosts file must be reviewed and merged manually. Generated DNS requires independent checking and correct installed serial state before application.

Input and output protections

  • Strict supported naming/settings rules, decimal IPv4 normalisation and complete conflict indexes.
  • Bounded input, logical records/tokens, names, alias depth/work, diagnostics and output; see format contract.
  • Structured fatal errors clear that tab's output and prevent export. Edits disable export until regeneration.
  • Separate BIND grammar; quoted data is tokenised before records, and file includes/generated directives are blocked without I/O.
  • Offline iterative CNAME resolution detects loops/conflicts and warns about missing targets. No invented address or network resolution.
  • Text-only rendering and per-tab captured metadata. Download URLs/anchors are released and clipboard/download failures are handled.

The BIND importer validates its supported grammar and mappings, not every non-host RDATA field or DNSSEC property. IPv6 and arbitrary CIDR/classless delegation are unsupported. This is not a DNS authority, universal validator or lossless round-trip tool.

Verification and release controls

Follow SFW v1.1.0. Source/unit, targeted-security and independent BIND jobs gate deployment in the prepared workflow. The security wrapper fails on configured findings, unreadable/error reports or missed production JS files; its synthetic test verifies failure behaviour without printing matched credentials. Four local rules are intentionally limited; comprehensive SAST, full secret-history scanning and scanner dependency locking/auditing remain improvements.

Playwright CI remains disabled under the owner's standing EX-001. The optional local matrix could not execute because usable browser binaries were unavailable. Adapter tests do not establish browser/a11y compliance.

Block releases for exposed credentials, active exploitation, confirmed reachable Critical vulnerabilities and newly introduced reachable High findings. Triage other findings using severity, reachability, exposure, fix availability and legacy/new status. Record an owner, mitigation and review date rather than silently suppressing controls. No risk acceptance or public-release approval is granted by the agent.

No push, tag, permission change or deployment occurred. Hosted branch-required checks and environment rules remain unverified. See verification, issue status and threat model.

Reporting

Do not post confidential inventory, credentials or exploit details in public issues. Use GitHub private vulnerability reporting if the owner has enabled it; enabled status is unverified and no private contact channel is invented here.

There aren't any published security advisories