BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//txlf2026//talk//M8H3K9
BEGIN:VTIMEZONE
TZID:US/Central
BEGIN:STANDARD
DTSTART:20251107T000000
TZNAME:CST
TZOFFSETFROM:-0600
TZOFFSETTO:-0600
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20260308T030000
RDATE:20270314T030000
TZNAME:CDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20261101T020000
TZNAME:CST
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:When AI Agents Get Shell Access: Securing Agentic Workloads on Lin
 ux - Pragya Keshap\, Akhilesh Keshap
DTSTART;TZID=US/Central:20261107T160000
DTEND;TZID=US/Central:20261107T165000
DTSTAMP:20260930T133617Z
UID:pretalx-txlf2026-M8H3K9@pretalx.com
DESCRIPTION:AI agents are evolving from conversational assistants into sys
 tems that can execute shell commands\, modify files\, call APIs\, start co
 ntainers\, and interact with production infrastructure. Giving an agent th
 ese capabilities can unlock powerful automation—but it can also turn a b
 ad model decision\, prompt-injection attack\, or compromised tool response
  into a serious security incident.\n\nThis session explores how Linux secu
 rity primitives can be used to contain agentic workloads before they are t
 rusted with real systems.\n\nWe will build a practical threat model for an
  AI agent with access to the operating system and examine risks such as ex
 cessive privileges\, unsafe command execution\, credential exposure\, unre
 stricted network access\, resource exhaustion\, and accidental modificatio
 n of host files.\n\nThe session will then demonstrate how to reduce these 
 risks using rootless containers\, Linux namespaces\, cgroups\, capability 
 dropping\, seccomp profiles\, read-only filesystems\, AppArmor or SELinux 
 policies\, network restrictions\, ephemeral credentials\, and detailed aud
 it logging.\n\nA live demonstration will compare the same agent running in
  an overly permissive environment and in a hardened Linux sandbox. We will
  observe which dangerous operations succeed\, which are blocked\, and how 
 the security controls affect the agent’s behavior.\n\nAttendees will lea
 ve with a practical architecture and security checklist for running AI age
 nts that can safely interact with Linux systems\, containers\, APIs\, and 
 infrastructure.
LOCATION:Lil Tex
URL:https://pretalx.com/txlf2026/talk/M8H3K9/
END:VEVENT
END:VCALENDAR