diff --git a/DANNY_STOCKER_INFRAFABRIC_DOSSIER.md b/DANNY_STOCKER_INFRAFABRIC_DOSSIER.md
index 8c0af0c..7d5b475 100644
--- a/DANNY_STOCKER_INFRAFABRIC_DOSSIER.md
+++ b/DANNY_STOCKER_INFRAFABRIC_DOSSIER.md
@@ -73,6 +73,12 @@ Live user diagnostics pages remain OAuth‑protected for privacy. For external r
 - Verifier (single file): https://infrafabric.io/static/hosted/iftrace.py
 - Model-tier invariants (Dave-proof run, 15 traces): https://infrafabric.io/static/hosted/EMO_DAVE_PROOF_MODEL_COMPARE_20251222T164352Z.md
 
+**Not proven yet (explicit gaps this portfolio does not hide):**
+- Clinical efficacy / outcomes (requires clinician-led study design + evaluation)
+- Completeness beyond the backend witness boundary (edge-level witnessing)
+- Key custody + rotation + compromise response (audit-grade key management)
+- Public append-only transparency log / external anchoring SLOs
+
 The evidence index links per‑trace pages and the underlying downloadable bundles + SHA256 sidecars.
 
 **Example traces (public bundles):**
@@ -148,6 +154,8 @@ When either filter changes the final user-visible output, the trace records `bef
 - It proves the *stack* can enforce specific invariants (language + formatting) across these model tiers for these prompts, with auditable corrections when needed.
 - It does not prove the models are equivalent on clinical judgment, crisis handling, or long‑horizon reasoning. Those require separate validation and are intentionally not claimed here.
 
+**Early-stage signal (bounded):** for these enforced invariants, we observed no tier-dependent failures across the three tested model tiers in this run.
+
 **Economic implication (bounded claim):** once these invariants are enforceable by the stack, model choice becomes a routing problem (default smaller, escalate when TRIAGE demands). Any claimed cost multipliers depend on provider pricing and are not asserted here.
 
 ---
diff --git a/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION_FULL.md b/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION_FULL.md
index 2c2f13f..142ff84 100644
--- a/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION_FULL.md
+++ b/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION_FULL.md
@@ -83,6 +83,12 @@ Live user diagnostics pages remain OAuth‑protected for privacy. For external r
 - Verifier (single file): https://infrafabric.io/static/hosted/iftrace.py
 - Model-tier invariants (Dave-proof run, 15 traces): https://infrafabric.io/static/hosted/EMO_DAVE_PROOF_MODEL_COMPARE_20251222T164352Z.md
 
+**Not proven yet (explicit gaps this portfolio does not hide):**
+- Clinical efficacy / outcomes (requires clinician-led study design + evaluation)
+- Completeness beyond the backend witness boundary (edge-level witnessing)
+- Key custody + rotation + compromise response (audit-grade key management)
+- Public append-only transparency log / external anchoring SLOs
+
 The evidence index links per‑trace pages and the underlying downloadable bundles + SHA256 sidecars.
 
 **Example traces (public bundles):**
@@ -158,6 +164,8 @@ When either filter changes the final user-visible output, the trace records `bef
 - It proves the *stack* can enforce specific invariants (language + formatting) across these model tiers for these prompts, with auditable corrections when needed.
 - It does not prove the models are equivalent on clinical judgment, crisis handling, or long‑horizon reasoning. Those require separate validation and are intentionally not claimed here.
 
+**Early-stage signal (bounded):** for these enforced invariants, we observed no tier-dependent failures across the three tested model tiers in this run.
+
 **Economic implication (bounded claim):** once these invariants are enforceable by the stack, model choice becomes a routing problem (default smaller, escalate when TRIAGE demands). Any claimed cost multipliers depend on provider pricing and are not asserted here.
 
 ---
diff --git a/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION_FULL.md b/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION_FULL.md
index 979cdaf..caff93c 100644
--- a/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION_FULL.md
+++ b/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION_FULL.md
@@ -83,6 +83,12 @@ Live user diagnostics pages remain OAuth‑protected for privacy. For external r
 - Verifier (single file): https://infrafabric.io/static/hosted/iftrace.py
 - Model-tier invariants (Dave-proof run, 15 traces): https://infrafabric.io/static/hosted/EMO_DAVE_PROOF_MODEL_COMPARE_20251222T164352Z.md
 
+**Not proven yet (explicit gaps this portfolio does not hide):**
+- Clinical efficacy / outcomes (requires clinician-led study design + evaluation)
+- Completeness beyond the backend witness boundary (edge-level witnessing)
+- Key custody + rotation + compromise response (audit-grade key management)
+- Public append-only transparency log / external anchoring SLOs
+
 The evidence index links per‑trace pages and the underlying downloadable bundles + SHA256 sidecars.
 
 **Example traces (public bundles):**
@@ -158,6 +164,8 @@ When either filter changes the final user-visible output, the trace records `bef
 - It proves the *stack* can enforce specific invariants (language + formatting) across these model tiers for these prompts, with auditable corrections when needed.
 - It does not prove the models are equivalent on clinical judgment, crisis handling, or long‑horizon reasoning. Those require separate validation and are intentionally not claimed here.
 
+**Early-stage signal (bounded):** for these enforced invariants, we observed no tier-dependent failures across the three tested model tiers in this run.
+
 **Economic implication (bounded claim):** once these invariants are enforceable by the stack, model choice becomes a routing problem (default smaller, escalate when TRIAGE demands). Any claimed cost multipliers depend on provider pricing and are not asserted here.
 
 ---