docs(confidence-engine): close pre-tester security review

docs/current-project-state.md

@@ -287,6 +287,35 @@ The following material learnings are carried forward as durable context for safe
   User-selected/active investigation ownership must survive substantive ties and
   question-formulation rejection. (Already documented in `docs/current-handoff.md`.)
 
+## 11. Pre-Tester Input Security Review (closed)
+
+**Status:** closed — sufficient for controlled external-user testing. Not a general security audit, penetration test, or production-launch certification.
+
+### SQL injection
+- No raw request-driven SQL construction found; persistence uses Supabase/PostgREST query-builder boundaries.
+
+### XSS
+- Normal user-controlled text is React-escaped; no unsafe HTML sink found.
+
+### Reasoning error disclosure
+- Browser-facing reasoning errors sanitised (commit `2cb2d55`).
+
+### Focused input bound
+- 10,000-character server-side + UI boundary on focused investigation (commit `a6796c6`).
+
+### Investigation snapshot size envelope
+- Authenticated persistence lacks a whole-snapshot size ceiling. Not a pre-tester blocker — legitimate size/depth is unknown; monitor via metadata later without logging content.
+
+### Prompt injection
+- **Low risk under the current architecture.** Untrusted text can influence model reasoning but no evidence it gains application authority. No model-accessible arbitrary tools, DB targeting, auth control, or privileged side effects found. Output passes structured validation before application mutation. Not a pre-tester blocker.
+
+### Deferred non-security observations
+- Orchestrator update flow indentation/control-flow clarity deferred.
+- Investigation overview validation may accept unexpected extra fields.
+- Provider JSON recovery permissiveness deferred as future robustness review.
+
+---
+
 ### RTO learning from Experiments 14–17
 
 Since the handoff document was written, further learning has emerged from Return-to-Origin work (RTO.14–17):