diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6a4d5500..f521bb23 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -4,9 +4,36 @@ All notable changes to ForgeLM are documented here.
 
 ## [Unreleased]
 
-Review-absorption rounds 3 + 4 follow-up to v0.5.7. No public surface
-change, but training-pipeline UX, doctor robustness, and test-isolation
-guarantees improve.
+Review-absorption rounds 3 + 4 + 5 follow-up to v0.5.7. No public
+surface change, but training-pipeline UX, doctor robustness, and
+test-isolation guarantees improve. Round 5 closes the residual
+CodeRabbit findings against the round-3/4 commits.
+
+### Fixed (round 5 — CodeRabbit follow-up)
+
+- **`forgelm/cli/_training.py::_preflight_numpy_torch_abi`** — any
+  unexpected exception from the underlying probe (corrupted torch
+  install where `torch.__version__` raises `AttributeError`, etc.)
+  is now caught and converted into a structured
+  `abi_preflight_crashed` JSON envelope. Previously the raw Python
+  traceback would pre-empt the `--output-format json` contract that
+  every other CLI failure path honours. Exit code stays
+  `EXIT_TRAINING_ERROR` (= 2), matching the broken-ABI verdict so
+  CI/CD branching doesn't need to distinguish "ABI bad" from "ABI
+  probe died".
+- **`CLAUDE.md`** — exit-code contract was stated as `0/1/2/3/4`,
+  but the canonical table in `docs/standards/error-handling.md`
+  documented `0/1/2/3/4/5` since v0.5.5 (Phase 22 added
+  `EXIT_WIZARD_CANCELLED = 5`). CLAUDE.md now matches the actual
+  contract — the standard was right, the agent guidance was stale.
+- **`docs/roadmap/completed-phases.md::Phase 12` summary** — Tier 1
+  status line claimed a `[ingestion-secrets]` extra "via
+  `detect-secrets` with regex fallback", but that extra was never
+  published; only `[ingestion-pii-ml]` exists in `pyproject.toml`'s
+  extras surface. Reworded to record the historical plan accurately
+  ("`detect-secrets` integration was originally planned as a
+  `[ingestion-secrets]` extra but deferred — only
+  `[ingestion-pii-ml]` ultimately shipped"). Pure docs accuracy fix.
 
 ### Added
 
diff --git a/CLAUDE.md b/CLAUDE.md
index 61f86fea..99b2feae 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -104,7 +104,7 @@ These come from the standards documents; summarized here for quick reference:
 1. **Config-driven.** Behaviour is determined by validated YAML. No env-var sniffing for behaviour (only for secrets). No hardcoded feature flags.
 2. **Reliability before features.** Every new capability ships with tests, docs, and CI coverage. "I'll add tests later" = the PR is not ready.
 3. **Optional dependencies as extras.** Heavy deps (`bitsandbytes`, `unsloth`, `deepspeed`, `lm-eval`, `wandb`, `mergekit`) live under `[project.optional-dependencies]` and raise `ImportError` with an install hint when missing.
-4. **Exit codes are a public contract.** 0/1/2/3/4 — see [error-handling.md](docs/standards/error-handling.md). CI/CD pipelines depend on these.
+4. **Exit codes are a public contract.** 0/1/2/3/4/5 — see [error-handling.md](docs/standards/error-handling.md) for the full table (`0=success`, `1=config`, `2=training`, `3=eval-failure`, `4=awaiting-approval`, `5=wizard-cancelled`). CI/CD pipelines depend on these.
 5. **Append-only audit log.** Every decision gate emits a structured event. Never edit or delete entries.
 6. **No silent failures.** No bare `except:`, no `except Exception: pass`, no `|| true` in CI, no logging-and-swallowing for anything except explicitly-non-fatal paths (webhooks, cleanup).
 7. **Bilingual where it counts.** User-facing docs are EN + TR mirrors. Code, CLI output, logs, config keys are English only.
diff --git a/docs/roadmap/completed-phases.md b/docs/roadmap/completed-phases.md
index 2789ac3a..df6534b0 100644
--- a/docs/roadmap/completed-phases.md
+++ b/docs/roadmap/completed-phases.md
@@ -923,7 +923,7 @@ suite because it is wall-clock-noisy.
 > and 12.5 — see [releases.md](releases.md#v050-document-ingestion-data-curation-pipeline).
 > Version-label references below preserve the historical planning trail.
 >
-> **Status:** ✅ **Tier 1 DONE** — landed on `development` for the (originally) `v0.5.2` cycle. All five must-have tasks shipped: MinHash LSH dedup option (`[ingestion-scale]` extra via `datasketch`), markdown-aware splitter (`--strategy markdown`), code/secrets leakage tagger (`[ingestion-secrets]` extra via `detect-secrets` with regex fallback), heuristic quality filter (`--quality-filter`), DOCX/Markdown table preservation. Tier 2 (Presidio adapter, Croissant metadata) and Tier 3 (`--all-mask` composite, wizard "audit first" hook) are deferred to a follow-up **Phase 12.5** backlog file (analogous to [Phase 11.5 section above](#phase-115-ingestion-audit-polish)). Modules: [`forgelm/data_audit.py`](../../forgelm/data_audit/), [`forgelm/ingestion.py`](../../forgelm/ingestion.py), [`forgelm/cli.py`](../../forgelm/cli/); tests: [`tests/test_data_audit_phase12.py`](../../tests/test_data_audit_phase12.py), [`tests/test_ingestion_phase12.py`](../../tests/test_ingestion_phase12.py); CLI tests added in [`tests/test_cli_subcommands.py`](../../tests/test_cli_subcommands.py).
+> **Status:** ✅ **Tier 1 DONE** — landed on `development` for the (originally) `v0.5.2` cycle. All five must-have tasks shipped: MinHash LSH dedup option (`[ingestion-scale]` extra via `datasketch`), markdown-aware splitter (`--strategy markdown`), code/secrets leakage tagger (regex-based; `detect-secrets` integration was originally planned as a `[ingestion-secrets]` extra but deferred — only `[ingestion-pii-ml]` for Presidio NER ultimately shipped in `pyproject.toml`), heuristic quality filter (`--quality-filter`), DOCX/Markdown table preservation. Tier 2 (Presidio adapter, Croissant metadata) and Tier 3 (`--all-mask` composite, wizard "audit first" hook) are deferred to a follow-up **Phase 12.5** backlog file (analogous to [Phase 11.5 section above](#phase-115-ingestion-audit-polish)). Modules: [`forgelm/data_audit.py`](../../forgelm/data_audit/), [`forgelm/ingestion.py`](../../forgelm/ingestion.py), [`forgelm/cli.py`](../../forgelm/cli/); tests: [`tests/test_data_audit_phase12.py`](../../tests/test_data_audit_phase12.py), [`tests/test_ingestion_phase12.py`](../../tests/test_ingestion_phase12.py); CLI tests added in [`tests/test_cli_subcommands.py`](../../tests/test_cli_subcommands.py).
 
 > **Note:** Phase 11 + 11.5 built the ingestion / audit lineage (Phase 11 → `v0.5.0`, Phase 11.5 → `v0.5.1`); this phase moves the same lineage from **enterprise-acceptable** to **enterprise-competitive**.
 
diff --git a/forgelm/cli/_training.py b/forgelm/cli/_training.py
index ab470c01..016423aa 100644
--- a/forgelm/cli/_training.py
+++ b/forgelm/cli/_training.py
@@ -46,10 +46,46 @@ def _preflight_numpy_torch_abi(json_output: bool) -> None:
     command if the mismatch is present.  No-op on healthy platforms
     (Linux, Apple Silicon, fresh-install Intel Mac), so zero false
     positives.
+
+    Any unexpected exception from the probe itself (a corrupted torch
+    install where ``torch.__version__`` raises ``AttributeError``, for
+    instance) converts into a structured ``abi_preflight_crashed``
+    envelope so the operator never sees a raw Python traceback
+    pre-empt the JSON contract — matching the rest of the CLI's
+    "every failure path carries a JSON envelope" rule per
+    ``docs/standards/error-handling.md``.
     """
     from ._abi_check import ABI_BROKEN, compute_numpy_torch_abi_status, format_abi_remediation
 
-    status, torch_version, numpy_version = compute_numpy_torch_abi_status()
+    try:
+        status, torch_version, numpy_version = compute_numpy_torch_abi_status()
+    except Exception as e:  # noqa: BLE001 — see docstring above: preflight
+        # is a sentinel that must NEVER let an unexpected exception escape
+        # to the user as a raw Python traceback.  CodeRabbit round-5
+        # absorption: wrap the probe so the JSON envelope contract holds
+        # even when the underlying torch / numpy install is corrupted
+        # enough that ``compute_numpy_torch_abi_status`` itself raises.
+        _report_training_error(
+            json_output,
+            payload={
+                "success": False,
+                "error": "abi_preflight_crashed",
+                "exception_class": type(e).__name__,
+                "exception_message": str(e),
+                "remediation": (
+                    "The torch/NumPy ABI preflight crashed unexpectedly. This "
+                    "usually indicates a corrupted torch or numpy install. Run "
+                    "`forgelm doctor` for the full environment diagnostic."
+                ),
+            },
+            log_msg=(
+                f"ABI preflight crashed unexpectedly ({type(e).__name__}: {e}). Run `forgelm doctor` for diagnostics."
+            ),
+            exit_code=EXIT_TRAINING_ERROR,
+            with_traceback=True,
+        )
+        return
+
     if status != ABI_BROKEN:
         return
     remediation = format_abi_remediation(torch_version, numpy_version)
diff --git a/tests/test_abi_check.py b/tests/test_abi_check.py
index e14e00bc..0648af06 100644
--- a/tests/test_abi_check.py
+++ b/tests/test_abi_check.py
@@ -209,3 +209,45 @@ def test_no_op_for(self, status):
         result = (status, "2.4.0", "1.26.4")
         with patch.object(_abi_check, "compute_numpy_torch_abi_status", return_value=result):
             _training._preflight_numpy_torch_abi(json_output=False)
+
+
+class TestPreflightHandlesProbeCrash:
+    """A crash in ``compute_numpy_torch_abi_status`` itself (corrupted
+    torch where ``torch.__version__`` raises, theoretical numpy ABI
+    poisoning that breaks the import-time machinery, etc.) must be
+    converted into a structured exit so the operator never sees a
+    raw Python traceback pre-empt the JSON envelope contract.
+
+    Round-5 absorption of CodeRabbit's MAJOR finding against the
+    round-3/4 preflight wiring."""
+
+    def _exploding_status(self):
+        raise AttributeError("module 'torch' has no attribute '__version__'")
+
+    def test_crash_exits_with_training_error_code(self, caplog):
+        with patch.object(_abi_check, "compute_numpy_torch_abi_status", side_effect=self._exploding_status):
+            with caplog.at_level("ERROR"), pytest.raises(SystemExit) as exc_info:
+                _training._preflight_numpy_torch_abi(json_output=False)
+        # Exit-code contract: the preflight crash maps to
+        # EXIT_TRAINING_ERROR (= 2 in the public table) — same class
+        # as the broken-ABI verdict, so CI/CD doesn't need to
+        # distinguish "ABI bad" from "ABI probe died".
+        assert exc_info.value.code == 2
+        # Operator-visible log must mention the actionable next step.
+        assert "forgelm doctor" in caplog.text
+        assert "ABI preflight crashed" in caplog.text
+
+    def test_crash_emits_structured_json_envelope(self, capsys):
+        with patch.object(_abi_check, "compute_numpy_torch_abi_status", side_effect=self._exploding_status):
+            with pytest.raises(SystemExit):
+                _training._preflight_numpy_torch_abi(json_output=True)
+        captured = capsys.readouterr()
+        envelope = json.loads(captured.out)
+        # Lock the key set — CI/CD consumers that parse this path
+        # need a stable shape, distinct from the "numpy_torch_abi_
+        # mismatch" envelope so they can branch on root cause.
+        assert envelope["success"] is False
+        assert envelope["error"] == "abi_preflight_crashed"
+        assert envelope["exception_class"] == "AttributeError"
+        assert "__version__" in envelope["exception_message"]
+        assert "forgelm doctor" in envelope["remediation"]