Clarify external watchdog direction

Change-source: pulse-maintainer

docs/releases/RELEASE_NOTES_v6.4.0-rc.10.md

@@ -8,7 +8,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, delivery routes can filter by severity, and informational events remain distinct from warnings.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **More accurate host disk health** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host, while empty Unraid slots remain neutral and Proxmox-linked agents avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, Proxmox backup views, and alert timelines present more complete and consistent context across desktop and narrow layouts.
 - **Safer governed actions** - Docker action results now carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.
 

docs/releases/RELEASE_NOTES_v6.4.0-rc.11.md

@@ -13,7 +13,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, delivery routes can filter by severity, and informational events remain distinct from warnings.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **More accurate host disk health** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host, while empty Unraid slots remain neutral and Proxmox-linked agents avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, backup views, and alert timelines present fuller context, while phone Settings use a searchable index, sticky section title, compact controls, and full-size touch targets.
 - **Safer governed actions** - Docker action results now carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.
 

docs/releases/RELEASE_NOTES_v6.4.0-rc.12.md

@@ -17,7 +17,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, delivery routes can filter by severity, and informational events remain distinct from warnings.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **More accurate host disk health** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host, while empty Unraid slots remain neutral and Proxmox-linked agents avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, Proxmox backup views, and alert timelines present more complete and consistent context across desktop and narrow layouts.
 - **Phone-friendly Settings** - On narrow layouts, phone Settings use a searchable index, sticky section title, compact controls, and full-size touch targets.
 - **Safer governed actions** - Docker action results carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.

docs/releases/RELEASE_NOTES_v6.4.0-rc.13.md

@@ -16,7 +16,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, delivery routes can filter by severity, and informational events remain distinct from warnings.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **More accurate host disk health** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host, while empty Unraid slots remain neutral and Proxmox-linked agents avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, Proxmox backup views, and alert timelines present more complete and consistent context across desktop and narrow layouts.
 - **Phone-friendly Settings** - On narrow layouts, phone Settings use a searchable index, sticky section title, compact controls, and full-size touch targets.
 - **Safer governed actions** - Docker action results carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.

docs/releases/RELEASE_NOTES_v6.4.0-rc.7.md

@@ -8,7 +8,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, and delivery routes can filter by severity.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **Host disk policies** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host while Proxmox-linked agents continue to avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, Proxmox backup views, and alert timelines present more complete and consistent context across desktop and narrow layouts.
 - **Safer governed actions** - Docker action results now carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.
 

docs/releases/RELEASE_NOTES_v6.4.0-rc.8.md

@@ -8,7 +8,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, delivery routes can filter by severity, and informational events remain distinct from warnings.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **Host disk policies** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host while Proxmox-linked agents continue to avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, Proxmox backup views, and alert timelines present more complete and consistent context across desktop and narrow layouts.
 - **Safer governed actions** - Docker action results now carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.
 

docs/releases/RELEASE_NOTES_v6.4.0-rc.9.md

@@ -8,7 +8,7 @@
 - **Better notification control** - Alerts can be snoozed individually, maintenance can recur by scope, escalation repeats can target specific destinations, delivery routes can filter by severity, and informational events remain distinct from warnings.
 - **Earlier capacity warnings** - Rolling metric windows and predictive storage forecasts surface sustained pressure and likely exhaustion before a single threshold breach becomes an outage.
 - **More accurate host disk health** - SMART sector, media, endurance, spare, and CRC thresholds can be tuned per host, while empty Unraid slots remain neutral and Proxmox-linked agents avoid duplicate disk-risk alerts.
-- **External availability monitoring** - Dead-man checks can notify when an expected external signal stops arriving, with configuration and recovery state persisted.
+- **External monitoring for Pulse itself** - Pulse can send a health signal to a Healthchecks-compatible watchdog on another host, with configuration and recovery state persisted.
 - **Clearer infrastructure details** - Resource drawers, Proxmox backup views, and alert timelines present more complete and consistent context across desktop and narrow layouts.
 - **Safer governed actions** - Docker action results now carry independently observed post-action state, while deployment enrollment and credential changes persist atomically.
 

scripts/check_public_docs.py

@@ -32,10 +32,18 @@ MARKDOWN_LINK = re.compile(r"!?\[[^\]]*\]\(([^)]+)\)")
 HTML_ASSET = re.compile(r"(?:src|href)=[\"']([^\"']+)[\"']")
 HEADING = re.compile(r"^#{1,6}\s+(.+?)\s*#*\s*$", re.MULTILINE)
 
+MISLEADING_PUBLIC_CLAIMS = {
+    r"dead-man checks can notify when an expected external signal stops arriving": (
+        "dead-man direction is reversed; Pulse sends its own health signal to an "
+        "external watchdog"
+    ),
+}
+
 
 def public_markdown_files() -> list[Path]:
     files = [ROOT / name for name in ROOT_DOCS if (ROOT / name).exists()]
     files.extend(sorted((ROOT / "docs").glob("*.md")))
+    files.extend(sorted((ROOT / "docs" / "releases").glob("*.md")))
     files.extend(sorted((ROOT / "docs" / "i18n").glob("**/*.md")))
     files.extend(sorted((ROOT / ".github" / "ISSUE_TEMPLATE").glob("*.md")))
     return files
@@ -110,6 +118,19 @@ def check_links(files: list[Path]) -> list[str]:
     return errors
 
 
+def check_public_claims(files: list[Path]) -> list[str]:
+    """Reject product claims whose direction changes the advertised capability."""
+
+    errors: list[str] = []
+    for path in files:
+        text = path.read_text(encoding="utf-8", errors="replace")
+        display_path = path.relative_to(ROOT)
+        for pattern, label in MISLEADING_PUBLIC_CLAIMS.items():
+            if re.search(pattern, text, flags=re.IGNORECASE):
+                errors.append(f"{display_path}: {label}")
+    return errors
+
+
 def check_current_claims() -> list[str]:
     errors: list[str] = []
     retired_patterns = {
@@ -152,6 +173,7 @@ def check_current_claims() -> list[str]:
 def main() -> int:
     files = public_markdown_files()
     errors = check_links(files)
+    errors.extend(check_public_claims(files))
     errors.extend(check_current_claims())
 
     if errors:

scripts/tests/test_public_docs_claims.py

@@ -0,0 +1,59 @@
+#!/usr/bin/env python3
+"""Guard public capability claims that can reverse the product contract."""
+
+from __future__ import annotations
+
+import importlib.util
+from pathlib import Path
+import sys
+import tempfile
+import unittest
+from unittest import mock
+
+
+ROOT = Path(__file__).resolve().parents[2]
+SCRIPT = ROOT / "scripts" / "check_public_docs.py"
+SPEC = importlib.util.spec_from_file_location("check_public_docs", SCRIPT)
+assert SPEC is not None and SPEC.loader is not None
+public_docs = importlib.util.module_from_spec(SPEC)
+sys.modules[SPEC.name] = public_docs
+SPEC.loader.exec_module(public_docs)
+
+
+class PublicDocsClaimsTest(unittest.TestCase):
+    def test_release_notes_are_in_the_public_claim_surface(self) -> None:
+        files = public_docs.public_markdown_files()
+
+        self.assertIn(ROOT / "docs/releases/RELEASE_NOTES_v6.4.0-rc.7.md", files)
+
+    def test_rejects_reversed_dead_man_direction(self) -> None:
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            note = root / "note.md"
+            note.write_text(
+                "Dead-man checks can notify when an expected external signal "
+                "stops arriving.\n",
+                encoding="utf-8",
+            )
+            with mock.patch.object(public_docs, "ROOT", root):
+                errors = public_docs.check_public_claims([note])
+
+        self.assertEqual(len(errors), 1)
+        self.assertIn("dead-man direction is reversed", errors[0])
+
+    def test_accepts_outbound_watchdog_description(self) -> None:
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            note = root / "note.md"
+            note.write_text(
+                "Pulse sends its own health signal to a watchdog on another host.\n",
+                encoding="utf-8",
+            )
+            with mock.patch.object(public_docs, "ROOT", root):
+                errors = public_docs.check_public_claims([note])
+
+        self.assertEqual(errors, [])
+
+
+if __name__ == "__main__":
+    unittest.main()