Merge branch 'master' into krr_push

Author: Avi-Robusta

docs/configuration/holmesgpt/index.rst

@@ -311,6 +311,38 @@ Reading the Robusta UI Token from a secret in HolmesGPT
 
 Run a :ref:`Helm Upgrade <Simple Upgrade>` to apply the configuration.
 
+Enable Holmes in Slack in the Platform
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+1. **Go to** https://platform.robusta.dev/
+
+2. **Navigate to:**
+   **Settings** → **AI Assistant**
+
+.. image:: /images/Enabling_AI_in_slack.png
+    :width: 1000px
+
+3. **Enable Holmes** using the toggle.
+
+4. **Click** **Connect Slack Workspace** to authorize Holmes in your Slack workspace.
+
+5. **Use Holmes in Slack**
+
+   In any Slack channel or thread, tag Holmes using `@holmes` like::
+
+     @holmes can you look into this
+
+   Or ask natural language questions about a specific cluster. Examples::
+
+.. code-block:: bash
+     @holmes what apps are crashing in my `prod-cluster`
+     @holmes show me the CPU usage for the frontend deployment in `staging-cluster`
+     @holmes why is my alert firing on `eu-prod-atc-eks`?
+     @holmes investigate high memory usage in `dev-cluster`
+
+   Holmes will respond in the thread with insights and troubleshooting steps based on the specified cluster.
+
+
 Test Holmes Integration
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 

docs/images/Enabling_AI_in_slack.png

@@ -1,9 +1,9 @@
 dependencies:
 - name: kube-prometheus-stack
   repository: https://prometheus-community.github.io/helm-charts
-  version: 55.7.0
+  version: 72.0.1
 - name: holmes
   repository: https://robusta-charts.storage.googleapis.com
   version: 0.11.5
-digest: sha256:af349cab87675093f9ad81330f889c31298bc0e9f918108a83702f33ae43ea41
-generated: "2025-06-10T09:36:25.85188+03:00"
+digest: sha256:3e3d5f742ecee2e068c88557aec4abca395ef1e9e90faabceb2671f27fe8ce52
+generated: "2025-06-22T16:35:28.432498+03:00"

helm/robusta/Chart.lock

@@ -11,7 +11,7 @@ appVersion: 0.0.0
 
 dependencies:
   - name: kube-prometheus-stack
-    version: 55.7.0
+    version: 72.0.1
     condition: enablePrometheusStack
     repository: "https://prometheus-community.github.io/helm-charts"
   - name: holmes

helm/robusta/Chart.yaml

@@ -42,7 +42,14 @@
 from robusta.integrations.receiver import ActionRequestReceiver
 from robusta.runner.web_api import WebApi
 from robusta.utils.stack_tracer import StackTracer
+from robusta.core.model.env_vars import ROBUSTA_API_ENDPOINT
+from cachetools import TTLCache
 
+RUNNER_GET_HOLMES_SLACKBOT_INFO = f"{ROBUSTA_API_ENDPOINT}/api/holmes/integrations/slack/runner"
+HOLMES_SLACKBOT_CACHE_TTL = int(os.getenv("HOLMES_SLACKBOT_CACHE_TTL", 15 * 60))
+
+# Define the cache with a single slot and the configured TTL
+_holmes_slackbot_cache = TTLCache(maxsize=1, ttl=HOLMES_SLACKBOT_CACHE_TTL)
 
 class RobustaSink(SinkBase, EventHandler):
     services_publish_lock = threading.Lock()
@@ -703,3 +710,26 @@ def __update_job(self, new_job: Job, operation: K8sOperationType):
                 self.__safe_delete_job(job_key)
                 self.__discovery_metrics.on_jobs_updated(1)
                 return
+
+    def is_holmes_slackbot_connected(self) -> bool:
+        if 'status' in _holmes_slackbot_cache:
+            return _holmes_slackbot_cache['status']
+        session_token = self.dal.get_session_token()
+        try:
+            message_json = {
+                "session_token": session_token,
+                "account_id": self.account_id,
+            }
+            response = requests.post(
+                RUNNER_GET_HOLMES_SLACKBOT_INFO,
+                json=message_json,
+                headers={"Content-Type": "application/json"}
+            )
+            response.raise_for_status()
+            is_connected = bool(response.json().get("integrations"))
+            _holmes_slackbot_cache['status'] = is_connected
+            return is_connected
+        except Exception as e:
+            logging.warning(f"Failed to get holmes slackbot info {e}")
+            _holmes_slackbot_cache['status'] = False
+            return False

helm/robusta/charts/kube-prometheus-stack-55.7.0.tgz

@@ -16,7 +16,7 @@ def __init__(self, sink_config: SlackSinkConfigWrapper, registry, is_preview=Fal
         self.slack_channel = slack_sink_params.slack_channel
         self.api_key = slack_sink_params.api_key
         self.slack_sender = slack_module.SlackSender(
-            self.api_key, self.account_id, self.cluster_name, self.signing_key, self.slack_channel, is_preview
+            self.api_key, self.account_id, self.cluster_name, self.signing_key, self.slack_channel, registry, is_preview
         )
         self.registry.subscribe("replace_callback_with_string", self)
 

helm/robusta/charts/kube-prometheus-stack-72.0.1.tgz

@@ -48,7 +48,6 @@
 from robusta.core.sinks.sink_base import KeyT
 from robusta.core.sinks.slack.slack_sink_params import SlackSinkParams
 from robusta.core.sinks.slack.preview.slack_sink_preview_params import SlackSinkPreviewParams
-
 from robusta.core.sinks.transformer import Transformer
 
 ACTION_TRIGGER_PLAYBOOK = "trigger_playbook"
@@ -62,7 +61,7 @@ class SlackSender:
     verified_api_tokens: Set[str] = set()
     channel_name_to_id = {}
 
-    def __init__(self, slack_token: str, account_id: str, cluster_name: str, signing_key: str, slack_channel: str, is_preview: bool = False):
+    def __init__(self, slack_token: str, account_id: str, cluster_name: str, signing_key: str, slack_channel: str, registry, is_preview: bool = False):
         """
         Connect to Slack and verify that the Slack token is valid.
         Return True on success, False on failure
@@ -80,6 +79,7 @@ def __init__(self, slack_token: str, account_id: str, cluster_name: str, signing
             timeout=SLACK_REQUEST_TIMEOUT,
             retry_handlers=all_builtin_retry_handlers(),
         )
+        self.registry = registry
         self.signing_key = signing_key
         self.account_id = account_id
         self.cluster_name = cluster_name
@@ -394,33 +394,6 @@ def __limit_labels_size(self, labels: dict, max_size: int = 1000) -> dict:
 
         return limited_labels
 
-    def __create_holmes_callback(self, finding: Finding) -> CallbackBlock:
-        resource = ResourceInfo(
-            name=finding.subject.name if finding.subject.name else "",
-            namespace=finding.subject.namespace,
-            kind=finding.subject.subject_type.value if finding.subject.subject_type.value else "",
-            node=finding.subject.node,
-            container=finding.subject.container,
-        )
-
-        context: Dict[str, Any] = {
-            "robusta_issue_id": str(finding.id),
-            "issue_type": finding.aggregation_key,
-            "source": finding.source.name,
-            "labels": self.__limit_labels_size(labels=finding.subject.labels),
-        }
-
-        return CallbackBlock(
-            {
-                "Ask HolmesGPT": CallbackChoice(
-                    action=ask_holmes,
-                    action_params=AIInvestigateParams(
-                        resource=resource, investigation_type="issue", ask="Why is this alert firing?", context=context
-                    ),
-                )
-            }
-        )
-
     @staticmethod
     def extract_mentions(title) -> (str, str):
         mentions = MENTION_PATTERN.findall(title)
@@ -667,6 +640,16 @@ def send_holmes_analysis(
         except Exception:
             logging.exception(f"error sending message to slack. {title}")
 
+    def get_holmes_block(self, platform_enabled: bool, slackbot_enabled) -> Optional[MarkdownBlock]:
+        if not platform_enabled and not slackbot_enabled:
+            return MarkdownBlock("_Ask AI questions about this alert, by connecting <https://platform.robusta.dev/create-account|Robusta SaaS> and tagging @holmes._")
+        elif platform_enabled and not slackbot_enabled:
+            return MarkdownBlock("_Ask AI questions about this alert, by adding @holmes to your <https://docs.robusta.dev/master/configuration/holmesgpt/index.html#enable-holmes-in-slack-in-the-platform|Slack>._")
+        elif platform_enabled and slackbot_enabled:
+            return MarkdownBlock("_Ask AI questions about this alert, by tagging @holmes in a threaded reply_")
+        return None
+
+
     def send_finding_to_slack(
         self,
         finding: Finding,
@@ -691,6 +674,15 @@ def send_finding_to_slack(
             thread_ts=thread_ts
         )
 
+    def __is_holmes_slackbot_enabled(self) -> bool:
+        robusta_sinks = self.registry.get_sinks().get_robusta_sinks() if self.registry else None
+        if not robusta_sinks:
+            logging.debug("No robusta sinks found, holmes not connected to slackbot")
+            return False
+
+        robusta_sink = robusta_sinks[0]
+        return robusta_sink.is_holmes_slackbot_connected()
+
     def __send_finding_to_slack(
         self,
         finding: Finding,
@@ -725,9 +717,6 @@ def __send_finding_to_slack(
         )
         blocks.append(links_block)
 
-        if HOLMES_ENABLED and HOLMES_ASK_SLACK_BUTTON_ENABLED:
-            blocks.append(self.__create_holmes_callback(finding))
-
         blocks.append(MarkdownBlock(text=f"*Source:* `{self.cluster_name}`"))
         if finding.description:
             if finding.source == FindingSource.PROMETHEUS:
@@ -753,6 +742,12 @@ def __send_finding_to_slack(
 
         blocks.append(DividerBlock())
 
+        is_holmes_slackbot_enabled = self.__is_holmes_slackbot_enabled()
+        holmes_block = self.get_holmes_block(platform_enabled, is_holmes_slackbot_enabled)
+        if holmes_block:
+            blocks.append(holmes_block)
+
+
         if len(attachment_blocks):
             attachment_blocks.append(DividerBlock())
 

src/robusta/core/sinks/robusta/robusta_sink.py

@@ -442,6 +442,7 @@ def main():
         cluster_name="test-cluster",
         signing_key="test-signing-key",
         slack_channel=SLACK_CHANNEL,
+        registry=None,
         is_preview=False
     )
 
@@ -452,6 +453,7 @@ def main():
         cluster_name="test-cluster",
         signing_key="test-signing-key",
         slack_channel=SLACK_CHANNEL,
+        registry=None,
         is_preview=True
     )