[SECURITY REVIEW] Try to break the v3.4 verification-boundary separation

[blackmore-technology-group]

Why this issue exists

ENTITY v3.4.0 retains the evidence/reality boundary introduced in v3.3 while adding the Global Passport, composable profile stack and domain-package surfaces.

That makes the same separation even more important:

1. cryptographic verification;
2. protocol verification;
3. evidence supporting a claim about the external world.

We want an unrelated external reviewer to try to falsify that separation on the current protected v3.4.0 release.

This is intentionally a bounded review task. You do not need to audit all of ENTITY.

Review target

Release: ENTITY v3.4.0 — Global Passport & Continuous Provenance

Protected release commit:

2db5bff64507b8d67642122a5ff2fc73dfef9152

v3.4 Global Passport sealed-kit SHA-256:

5869a3fd0ed6cb9f65bf4b20c3bd64933cad82f4aef05c5809e2e05af921f230

Global Passport schema SHA-256:

4fbfed9be1b1484bc5d28b8101d1c908b2ccced13e4e99ec896c5b054892ebdd

Canonical v3.4 campaign result SHA-256:

ac7504cce70576008cff069607619660a4b9bf0cad43b3f3de81078f1e80d9ba

Start here:

• Developer portal: DEVELOPERS.md
• v3.4 documentation: docs/v3.4/README.md
• evidence/reality implementation: src/37_Verifiable_Reality/
• Global Passport implementation: src/38_Global_Passports/
• engineering evidence: docs/ENGINEERING_EVIDENCE.md
• security program/material: SECURITY.md and docs/security/

Pick one boundary to attack

A useful review can focus on just one of these:

• signature → truth escalation — can a cryptographically valid record acquire a stronger external-truth state than the evidence warrants?
• protocol validity → external authority escalation — can a valid transition silently imply authority outside its granted scope?
• external anchor → sovereignty escalation — can a registry/API/sensor/attester gain general ENTITY authority merely because it supplied evidence?
• provenance → rights escalation — can origin evidence become ownership/licensing/usage authority without the required rights chain?
• attestation scope escape — can an attester act outside its permitted class/scope or remain effective after revocation?
• Global Passport/profile escalation — can profile composition or standards mapping accidentally strengthen truth, rights or authority?
• dispute/supersession failure — can a disputed or superseded claim continue to produce authoritative downstream state incorrectly?
• causal-attribution overclaim — can correlation or temporal proximity be promoted into a supported economic-causality edge without the required evidence?

Deliverable

Any one of the following is useful:

• a reproducible counterexample;
• a failing test/vector;
• a short threat model identifying an exploitable transition;
• a specification ambiguity with two plausible incompatible interpretations;
• a concise review explaining why the chosen boundary appears to hold, including what you actually tested.

Please identify the exact target commit, files/modules reviewed, method, reproduction steps/test code where safe, expected vs observed behavior, classification of the finding, and important exclusions.

Security-sensitive findings

If you find something exploitable, do not publish exploit details, keys, credentials or private data in this issue. Use the repository's private vulnerability reporting path in SECURITY.md and reference this issue.

Non-sensitive counterexamples and architectural criticism are welcome here.

What counts as success

A pass is useful. A failure is useful. An ambiguity is useful.

The goal is independently authored evidence that another engineer can understand and, where applicable, reproduce. BTG-controlled testing is not treated as independent external validation.

This issue does not promise a monetary bounty, certification engagement or paid audit; any commercial review arrangement would be separate.