diff --git a/scripts/custody-describe.py b/scripts/custody-describe.py
index 0e8f210..ea1cff3 100755
--- a/scripts/custody-describe.py
+++ b/scripts/custody-describe.py
@@ -12,7 +12,7 @@ because a confidently wrong `used_by` is worse than an empty one.
 
 `apply` writes custom_metadata only (`bao kv metadata patch`), for entries
 marked `review: accepted` with all five fields filled. It never reads or writes
-a value, prints nothing but paths and booleans, and runs under the founder's
+a value, prints nothing (results go to a 0600 receipt), and runs under the founder's
 admin session (warden access openbao-platform-admin-login --exec), since
 ops-mason-survey cannot write metadata.
 """
@@ -121,20 +121,31 @@ def check(doc: dict) -> int:
     return 0 if bad == 0 else 1
 
 
-def apply(doc: dict) -> int:
-    failed = 0
+RECEIPT = os.path.join(HOME, ".local/state/railiance/receipts/custody-describe.json")
+
+
+def apply(doc: dict, receipt: str = RECEIPT) -> int:
+    """Silent: the warden wrapper refuses any output, so results go to a 0600 receipt."""
+    import json
+    results = {}
     for path, e in doc.items():
         if e.get("review") != "accepted" or any(not e.get(k) for k in REQUIRED):
-            print(f"skip {path}")
+            results[path] = "skipped"
             continue
         mount, rest = path.split("/", 1)
         args = ["bao", "kv", "metadata", "patch", f"-mount={mount}"]
         args += [f"-custom-metadata={k}={_clip(e[k])}" for k in REQUIRED]
         args.append(rest)
         ok = subprocess.run(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL).returncode == 0
-        failed += not ok
-        print(f"{'patched' if ok else 'FAILED '} {path}")
-    return 1 if failed else 0
+        results[path] = "patched" if ok else "failed"
+    os.makedirs(os.path.dirname(receipt), mode=0o700, exist_ok=True)
+    old_umask = os.umask(0o077)
+    try:
+        with open(receipt, "w") as f:
+            json.dump(results, f, indent=1, sort_keys=True)
+    finally:
+        os.umask(old_umask)
+    return 1 if "failed" in results.values() else 0
 
 
 def main(argv: list[str]) -> int:
diff --git a/tests/test_custody_describe.py b/tests/test_custody_describe.py
index 6591793..220cae8 100644
--- a/tests/test_custody_describe.py
+++ b/tests/test_custody_describe.py
@@ -9,14 +9,19 @@ _spec.loader.exec_module(cd)
 FULL = dict(description="d", owner="o", used_by="u", rotation="r", on_loss="l")
 
 
-def test_apply_patches_only_accepted_complete_entries_and_only_metadata():
+def test_apply_patches_only_accepted_complete_entries_and_only_metadata(tmp_path, capsys):
     doc = {
         "platform/workloads/a/b": {"review": "accepted", **FULL},
         "platform/workloads/c/d": {"review": "draft", **FULL},
         "operators/e": {"review": "accepted", **dict(FULL, used_by="")},
     }
     with patch.object(cd.subprocess, "run", return_value=MagicMock(returncode=0)) as run:
-        assert cd.apply(doc) == 0
+        assert cd.apply(doc, receipt=str(tmp_path / 'r.json')) == 0
+    assert capsys.readouterr() == ('', '')
+    import json, os, stat
+    r = tmp_path / 'r.json'
+    assert json.load(open(r)) == {'platform/workloads/a/b': 'patched', 'platform/workloads/c/d': 'skipped', 'operators/e': 'skipped'}
+    assert stat.S_IMODE(os.stat(r).st_mode) == 0o600
     assert run.call_count == 1
     args = run.call_args.args[0]
     assert args[:5] == ["bao", "kv", "metadata", "patch", "-mount=platform"]
@@ -25,10 +30,11 @@ def test_apply_patches_only_accepted_complete_entries_and_only_metadata():
     assert run.call_args.kwargs["stdout"] is cd.subprocess.DEVNULL and run.call_args.kwargs["stderr"] is cd.subprocess.DEVNULL
 
 
-def test_apply_reports_failure_without_output():
+def test_apply_reports_failure_without_output(tmp_path, capsys):
     doc = {"platform/workloads/a/b": {"review": "accepted", **FULL}}
     with patch.object(cd.subprocess, "run", return_value=MagicMock(returncode=2)):
-        assert cd.apply(doc) == 1
+        assert cd.apply(doc, receipt=str(tmp_path / 'r.json')) == 1
+    assert capsys.readouterr() == ('', '')
 
 
 def test_check_refuses_drafts_missing_and_overlong_fields():
diff --git a/workplans/MASON-WP-0004-credential-inventory-descriptions.md b/workplans/MASON-WP-0004-credential-inventory-descriptions.md
index e1d0ee0..9564532 100644
--- a/workplans/MASON-WP-0004-credential-inventory-descriptions.md
+++ b/workplans/MASON-WP-0004-credential-inventory-descriptions.md
@@ -115,9 +115,12 @@ have gaps left empty on purpose (nextcloud x2, keycape-factor-issuer,
 grafana-admin, smtp, and four `used_by`). Every entry is `review: draft`.
 `apply` writes custom_metadata only, for `review: accepted` entries, silently
 (`tests/test_custody_describe.py`, suite 157 passed). Founder run, after
-review: `warden access openbao-platform-admin-login --exec -- python3
+review: `warden access openbao-platform-admin-login --exec -- sh -c 'python3
 /home/worsch/ops-mason/scripts/custody-describe.py apply
-/home/worsch/ops-mason/manifests/custody-descriptions.yaml`.
+/home/worsch/ops-mason/manifests/custody-descriptions.yaml >/dev/null 2>&1'`,
+then read `~/.local/state/railiance/receipts/custody-describe.json`.
+(First attempt 2026-10-09 refused by the wrapper because apply printed status
+lines; nothing written, all entries were still drafts. apply is now silent.)
 
 ## Resolve the rest with their owners