| ID | Risk | Severity | Current status | Guard | Upstream |
|---|---|---|---|---|---|
| TUI-R-001 | Explicit root overwritten by saved state. | High | Mitigated; explicit root wins. | Explicit-root sentinel and path tests. | risk-register-and-threat-model-2026-06-02.md (state durability) |
| TUI-R-002 | Cost display diverges from run cost. | High | Mitigated; shared controller ledger is active. Provider usage availability remains an external limit. | Active-path cost and budget tests. | risk-register-and-threat-model-2026-06-02.md (CG-03 cost accounting; DS-01) |
| TUI-R-003 | TUI undo differs from REPL undo. | High | Mitigated; both use the run undo journal and TUI has no global fallback. | tests/tui/test_tui_undo_scope.py. |
risk-register-and-threat-model-2026-06-02.md (DS-05 undo semantics) |
| TUI-R-004 | TUI chat bypasses controller semantics. | Medium-high | Mitigated; task execution delegates to ChatSessionController. |
Surface-parity and command-path tests. | phase-0-trust-repair-risk-brief-2026-06-04.md |
| TUI-R-005 | Tests assert helper state only. | Medium-high | Mitigated for known daily-driver paths; monitor new commands. | Headless command-path fixtures. | phase-0-trust-repair-risk-brief-2026-06-04.md (test integrity) |
| TUI-R-006 | Approval prompt hides scope. | High | Mitigated; empty path scope is rejected. | Approval visibility and empty-scope tests. | risk-register-and-threat-model-2026-06-02.md (approval visibility; DS-12) |
Review this file whenever teaagent/tui/, teaagent/chat_session_controller.py, or TUI
approval routing changes.