<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/dd62c63a3c514aa9abc21dfca17d715c&quot; frameborder=&quot;0&quot; width=&quot;1920&quot; height=&quot;1440&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1440</height><width>1920</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1440</thumbnail_height><thumbnail_width>1920</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/dd62c63a3c514aa9abc21dfca17d715c-1a935d8dfb378867.gif</thumbnail_url><duration>352.897</duration><title>AI Read Only Kubernetes Incident Triage Plugin</title><description>This Loom demonstrates an Oncall and Triage plugin for early Kubernetes incident triage with a read-only safety model. It shows a Cloud Code run for a Cloud Cluster with Elevated Errors in the last 30 minutes, correlating metrics and error logs and generating hypotheses and next steps. The author emphasizes safety decisions: exposing only read-only tools like Search Logs and Cluster Overview to reduce injection and using a hook as a hard stop against destructive commands even if a user confirms. It also covers handling a “silent failure” warning, where the skill adjusts for invalid parameters such as requesting hours equals one but defaulting to the actual 30 minute data window. Finally, it previews a customer guide including a persona table and build steps for skills, agents, and hooks.</description></oembed>