diff --git a/puppy_kit/mcp/server.py b/puppy_kit/mcp/server.py
index 9c816e3..e68b6c5 100644
--- a/puppy_kit/mcp/server.py
+++ b/puppy_kit/mcp/server.py
@@ -118,7 +118,8 @@ def dd_incidents_get(incident_id: str) -> str:
     (bug report content, user scope, what actually triggered the incident).
 
     The fields block contains triage data: root_cause, summary, triage_findings,
-    services, teams, needshumanattention, triagecompleted, githubreferences, etc.
+    services, teams, needshumanattention, triagecompleted, NeedsWrenchBotCheck,
+    githubreferences, etc.
 
     If timeline_truncated is true, the incident has more than 10 cells —
     call dd_incidents_get_timeline to read the full timeline.
@@ -150,7 +151,7 @@ def dd_incidents_create(
     Use this when a new problem is detected and requires an incident record. Requires
     a descriptive title. Severity and team are optional and can be set via dd_incidents_update
     after creation. Returns the created incident ID which can then be passed to
-    dd_incidents_update to set status and add detail.
+    dd_incidents_update to add detail. To set status use dd_incidents_set_status.
 
     Args:
         title: Short descriptive title of the issue.
@@ -193,6 +194,7 @@ def dd_incidents_update(
     needs_human_attention: str | None = None,
     triage_completed: str | None = None,
     is_duplicate: str | None = None,
+    needs_wrench_bot_check: str | None = None,
     github_refs: str | None = None,
     datadog_refs: str | None = None,
     teams: list[str] | None = None,
@@ -201,26 +203,29 @@ def dd_incidents_update(
 ) -> str:
     """Update an existing Datadog incident with title, severity, commander, and custom fields.
 
-    Use to progress an incident and populate custom workflow fields. At least one field
-    must be provided. Use custom fields to document triage results, root cause, detection
-    method, and related resources. To change incident status use dd_incidents_set_status.
+    Use to populate triage results, root cause, detection method, and related resources.
+    At least one field must be provided.
+
+    To change incident status (active/stable/resolved) use dd_incidents_set_status — not
+    this tool. Passing status here will have no effect.
 
     Args:
         incident_id: The incident UUID to update.
         title: New incident title (optional).
         severity: New severity — SEV-1 through SEV-5 (optional).
         assignee: Assignee name (e.g., 'muhammad', 'willem', 'jeong') to set as incident commander (optional).
-        summary: Brief summary of the incident (optional).
-        root_cause: Root cause description (optional).
-        triage_findings: Triage findings (optional).
+        summary: Brief one-line summary of who is affected and what they cannot do (optional).
+        root_cause: Root cause description — specific component, error class, and code path (optional).
+        triage_findings: Evidence log — counts, rates, blast radius, GitHub findings (optional).
         detection_method: How the incident was detected: 'monitor', 'employee', 'customer',
                           'alert', or 'unknown' (optional).
-        needs_monitoring: 'yes' or 'no' (optional).
-        needs_human_attention: 'yes' or 'no' (optional).
-        triage_completed: 'yes' or 'no' (optional).
-        is_duplicate: 'yes' or 'no' (optional).
-        github_refs: GitHub references (optional).
-        datadog_refs: Datadog references (optional).
+        needs_monitoring: 'yes' or 'no' — whether the incident needs ongoing monitoring (optional).
+        needs_human_attention: 'yes' or 'no' — whether a human must take action (optional).
+        triage_completed: 'yes' or 'no' — set to 'yes' when triage is done (optional).
+        is_duplicate: 'yes' or 'no' — set to 'yes' if this duplicates another incident (optional).
+        needs_wrench_bot_check: 'yes' or 'no' — setting to 'yes' triggers the triage webhook (optional).
+        github_refs: GitHub PR or issue URLs (optional).
+        datadog_refs: Related Datadog URLs (optional).
         teams: List of team names (optional).
         services: List of service names (optional).
         related_incidents: List of related incident IDs (optional).
@@ -252,6 +257,8 @@ def dd_incidents_update(
         args += ["--triage-completed", triage_completed]
     if is_duplicate is not None:
         args += ["--is-duplicate", is_duplicate]
+    if needs_wrench_bot_check is not None:
+        args += ["--needs-wrench-bot-check", needs_wrench_bot_check]
     if github_refs is not None:
         args += ["--github-refs", github_refs]
     if datadog_refs is not None:
@@ -350,7 +357,7 @@ def dd_incidents_get_timeline(incident_id: str) -> str:
 def dd_incidents_delete(incident_id: str) -> str:
     """Permanently delete a Datadog incident. Use only to remove erroneous or duplicate records.
 
-    This is irreversible. For closing a real incident, use dd_incidents_update with
+    This is irreversible. For closing a real incident, use dd_incidents_set_status with
     status='resolved' instead — that preserves the incident history. Only call this
     when an incident was created by mistake and should not exist at all.