<?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/96058cb46f264c20a8092d029f43c522&quot; frameborder=&quot;0&quot; width=&quot;1114&quot; height=&quot;835&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>835</height><width>1114</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>835</thumbnail_height><thumbnail_width>1114</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/96058cb46f264c20a8092d029f43c522-463107df557db822.gif</thumbnail_url><duration>76.255</duration><title>Team Beacon: CLI for Risky Agent Behaviour</title><description>This Loom explains how Team Beacon’s CLI helps analyze risky behavior from coding agents that run with inherited access to a user’s identity. Justin and Sovit describe the problem that agents can use your SSH keys, GitHub authentication, cloud credentials, and access your machine or the open network. In a live demo, they kick off a cloud session where an agent is asked to describe a repository and reverse engineer its architecture. The CLI captures and normalizes agent telemetry, applies detections and analysis to the agent’s trajectory, and then shows the results in their dashboard.</description></oembed>