Investigations

investigations serves as your primary workspace for diagnosing issues, exploring anomalies, and uncovering insights you can ask questions across metrics, logs, and traces, and iteratively troubleshoot problems in a conversational flow when you land on the investigations page, the main area presents three quick access cards to help you get started previous investigations opens the full list of past investigations so you can find and revisit completed or in progress rca reports integration hub takes you to the sdk and integration guide for connecting external systems to the assistant skills opens the skills configuration page where you can create and edit agent skills adding context to a message the + button in the chat input bar opens a context picker that lets you attach relevant resources to your message before sending it this helps the assistant understand the scope of your query without requiring you to describe everything from scratch select a category on the left panel, then browse the available resources on the right and attach those relevant to your investigation available context categories include services attach apm instrumented services you want the assistant to focus on dashboards link workspace dashboards to provide visual context alarms include alert rules relevant to your query k8s resources reference kubernetes clusters, namespaces, deployments, and pods synthetic checks attach uptime or browser monitors saved queries include saved log, trace, or metric queries to anchor the investigation ​​​​​​​​​​​​​​​​ investigations list clicking previous investigations from the main investigations page opens the full investigations list this shows all past and active investigations in your workspace each row in the list displays the following information title a brief summary or auto generated description of the investigation status the current state of the investigation, typically shown as completed or in progress time range the observation window that was used when the investigation was run created the date and time the investigation was initiated created by the user who triggered the investigation starting a new investigation to start a new investigation, click the new investigation button in the top right corner of the investigations page this brings you to the investigations chat interface where you can describe the issue or symptom you want analyzed in the input bar at the bottom of the screen the kloudmate investigator agent will collect your brief and begin a background rca workflow ​​​​​​​​​​​​​​​​ viewing an investigation click any investigation title to open the full investigation report the report is structured to give you a clear picture of what was analyzed, what was found, and what the likely causes are at the top of the report, the investigation context panel summarizes the scope of the analysis, including the services or resources examined, the time range observed, and the symptom or trigger that was described it also shows the number of ai credits consumed for the investigation and a status badge indicating whether it is completed or still running if the outcome could not be determined, the badge will show unknown below the context panel, the identified user impact section provides a plain language summary of how the issue may be affecting end users or downstream systems this is designed to give a quick understanding of the situation before diving into the technical details the main body of the report contains the active hypotheses section, which presents a ranked list of potential root causes uncovered during the investigation each hypothesis includes the following hypothesis name a concise label for the suspected root cause severity indicates the likely impact level high, medium, or low description a detailed explanation of why this hypothesis was raised, what it implies for the system, and how it relates to the observed symptom supporting evidence a list of data points from traces, topology, logs, and metrics that validate the hypothesis ​​​​​​​​​​​​​​​​ relevant resources tags that link to the specific services, pods, or other entities implicated in the hypothesis, making it easy to navigate directly to the affected components