--- title: Why sandboxing your agent is not enough | daily.dev description: Sandboxing AI agents in Kubernetes provides security and isolation, but it's not enough for large-scale deployments. Two projects are compared: agent-sandbox,... canonical: https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Why sandboxing your agent is not enough | daily.dev og:description: Sandboxing AI agents in Kubernetes provides security and isolation, but it's not enough for large-scale deployments. Two projects are compared: agent-sandbox,... og:url: https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk og:image: https://api.daily.dev/og/posts/M5nC5KDyK.png og:image:alt: Why sandboxing your agent is not enough og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Why sandboxing your agent is not enough **[CNCF](https://daily.dev/sources/cncf)** · 5 min read · 13 upvotes · 0 comments ## Summary Sandboxing AI agents in Kubernetes provides security and isolation, but it's not enough for large-scale deployments. Two projects are compared: agent-sandbox, which focuses on secure, Kubernetes-native execution with strong identities and lifecycle management, and agent-substrate, which adds density, efficiency, and dynamic on-demand execution. Agent-substrate decouples agent actors from pod lifecycles, letting agents run in short bursts on shared worker pools and suspend when idle — similar to serverless workloads. This approach allows hundreds of agents to coexist on minimal infrastructure using lightweight runtimes like gVisor or Kata Containers. The author demonstrates integrating agent-substrate with kagent, where six agents share a single worker pod, scaling horizontally only when concurrency demands it. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the difference between agent-sandbox and agent-substrate for running AI agents on Kubernetes? Agent-sandbox is a Kubernetes SIG Apps project providing a Sandbox CRD and controller focused on secure identities, persistent storage, and lifecycle management for agent pods. Agent-substrate is a separate standalone project built on top of Kubernetes that goes further, enabling agents to run in short-lived worker pods, suspend when idle, and resume on demand for higher density, lower latency, and better resource efficiency at scale. _Teams weighing Kubernetes agent runtimes can follow how these projects evolve on daily.dev._ ### How does agent-substrate reduce idle resource consumption for AI agents running on Kubernetes? Agent-substrate decouples the agent actor's lifecycle from the underlying worker pod, letting agents execute in short bursts, suspend when idle, and resume later on any available worker pod. This means many agents can share a single worker pool instead of each requiring a dedicated always-on pod, using lightweight sandbox runtimes like gVisor or Kata Containers, cutting idle resource waste while preserving isolation. _Engineers optimizing agent fleet costs can track efficiency techniques like this on daily.dev._ ## Similar posts on daily.dev - [Kubernetes won the container decade. Google’s Agent Substrate wants the next one.](https://daily.dev/posts/kubernetes-won-the-container-decade-google-s-agent-substrate-wants-the-next-one--htft1i3hx) · The New Stack · 3 upvotes · 0 comments - [Is a Pod the right deployment unit for an AI agent?](https://daily.dev/posts/is-a-pod-the-right-deployment-unit-for-an-ai-agent--ypylfpli4) · CNCF · 8 upvotes · 0 comments - [Pods as Workers, Not Agents: Rethinking the Deployment Unit for AI Agents on Kubernetes](https://daily.dev/posts/pods-as-workers-not-agents-rethinking-the-deployment-unit-for-ai-agents-on-kubernetes-iiclsmfiu) · InfoQ · 3 upvotes · 2 comments --- Tags: [#kubernetes](https://daily.dev/tags/kubernetes), [#ai-agents](https://daily.dev/tags/ai-agents), [#serverless](https://daily.dev/tags/serverless) [View this post on daily.dev](https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Why sandboxing your agent is not enough","url":"https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk"},"datePublished":"2026-07-07T11:32:23.403Z","dateModified":"2026-09-13T20:02:53.030Z","description":"Sandboxing AI agents in Kubernetes provides security and isolation, but it's not enough for large-scale deployments. Two projects are compared: agent-sandbox,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/73635257569d7d958b7eab1e0ef1689e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/73635257569d7d958b7eab1e0ef1689e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"CNCF","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"CNCF","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/55536a03390741eebd96d2f9b74d2d8d","url":"https://daily.dev/sources/cncf"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":13},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"kubernetes,ai-agents,serverless","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"CNCF","item":"https://daily.dev/sources/cncf"},{"@type":"ListItem","position":3,"name":"Why sandboxing your agent is not enough"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk#faq","mainEntity":[{"@type":"Question","name":"What is the difference between agent-sandbox and agent-substrate for running AI agents on Kubernetes?","acceptedAnswer":{"@type":"Answer","text":"Agent-sandbox is a Kubernetes SIG Apps project providing a Sandbox CRD and controller focused on secure identities, persistent storage, and lifecycle management for agent pods. Agent-substrate is a separate standalone project built on top of Kubernetes that goes further, enabling agents to run in short-lived worker pods, suspend when idle, and resume on demand for higher density, lower latency, and better resource efficiency at scale. Teams weighing Kubernetes agent runtimes can follow how these projects evolve on daily.dev."}},{"@type":"Question","name":"How does agent-substrate reduce idle resource consumption for AI agents running on Kubernetes?","acceptedAnswer":{"@type":"Answer","text":"Agent-substrate decouples the agent actor's lifecycle from the underlying worker pod, letting agents execute in short bursts, suspend when idle, and resume later on any available worker pod. This means many agents can share a single worker pool instead of each requiring a dedicated always-on pod, using lightweight sandbox runtimes like gVisor or Kata Containers, cutting idle resource waste while preserving isolation. Engineers optimizing agent fleet costs can track efficiency techniques like this on daily.dev."}}]} ```