Skip to content

Security Privacy and Limitations

Meridian Verity Group edited this page Jul 19, 2026 · 1 revision

Security, Privacy, and Limitations

Evaluation security posture

The artifact is designed to make authorization behavior falsifiable through deterministic examples and fail-closed negative cases. It models exact-effect binding, current policy state, bounded validity, scope, status/recency, authenticity, replay resistance, and optional downstream enforcement.

It is not production infrastructure and must not be treated as a complete production security architecture.

Security-relevant failure families

A verifier should deny when authorization-critical evidence is:

  • missing or unavailable under a policy that requires it;
  • stale, expired, revoked, or from an unacceptable epoch;
  • ambiguous or conflicting;
  • malformed or strictly typed incorrectly;
  • unsupported by the selected profile;
  • cryptographically unverifiable;
  • bound to a different action, target, audience, tenant, purpose, jurisdiction, operation, or budget;
  • replayed or involved in a failed atomic replay reservation; or
  • presented on a path that cannot establish the required gate or downstream capability.

Production assumptions not established here

A production deployment would still need to establish:

  • complete inventory and mediation of effect-bearing paths;
  • hardened interceptor placement and downstream enforcement;
  • root-of-trust, issuer, key, and certificate lifecycle;
  • durable replay state and partition/failure semantics;
  • policy authoring, approval, rollout, rollback, and emergency governance;
  • status and revocation availability targets;
  • secure audit storage, retention, access control, and redaction;
  • operational monitoring, incident response, recovery, and change control;
  • privacy impact assessment and data minimization; and
  • domain-specific legal, regulatory, safety, and risk controls.

The presence of tests or coverage does not prove these deployment properties.

Privacy and data handling

Use only synthetic or public-safe data. Do not place any of the following in the repository, wiki, public issues, logs, vectors, or screenshots:

  • customer or end-user data;
  • cardholder data, PAN, SAD, bank or payment credentials;
  • API keys, signing keys, tokens, passwords, or production identifiers;
  • production processor, wallet, issuer, or network details;
  • proprietary implementation mappings;
  • confidential legal or patent analysis;
  • private commercial or strategy material; or
  • third-party content you are not authorized to disclose.

Decision records should prefer digests and references over raw sensitive payloads and should include only what is needed for reproduction and audit.

Reporting a suspected vulnerability

Do not open a public issue containing exploit details or sensitive material.

Use GitHub's private vulnerability reporting route:

Report a vulnerability privately

If that route is not visible, disclose no technical details publicly. Send only a minimal request for a private contact channel through an available maintainer contact. The repository owner should keep SECURITY.md synchronized with the actual private route.

Disclosure scope

Appropriate reports include vulnerabilities in the public evaluation artifact, verifier logic, parsers, canonicalization, signature checks, replay handling, release-integrity tooling, or workflows. Production-system issues, real credentials, live payment data, or non-public deployments are outside the public artifact and must never be posted publicly.

Licensing and standards limitations

The evaluation-only license permits limited non-production evaluation, research, interoperability, and standards-discussion activity. It does not grant production or commercial rights, patent rights, certification rights, or a standards commitment. The Internet-Draft has no IETF consensus or endorsement.

Clone this wiki locally