--- title: AI Agent FULL Security Guide - Don't Get Hacked | daily.dev description: A comprehensive security guide for AI agent deployments covering three main attack surfaces: the server (VPS), prompt injection, and tool permissions. Covers... canonical: https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: AI Agent FULL Security Guide - Don't Get Hacked | daily.dev og:description: A comprehensive security guide for AI agent deployments covering three main attack surfaces: the server (VPS), prompt injection, and tool permissions. Covers... og:url: https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f og:image: https://api.daily.dev/og/posts/0i3TOxn7F.png og:image:alt: AI Agent FULL Security Guide - Don't Get Hacked 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. # AI Agent FULL Security Guide - Don't Get Hacked **[Tech With Tim](https://daily.dev/sources/TechWithTim)** · 38 min read · 1 upvotes · 0 comments ## Summary A comprehensive security guide for AI agent deployments covering three main attack surfaces: the server (VPS), prompt injection, and tool permissions. Covers setting up Tailscale as a private mesh VPN to isolate a VPS from the public internet, disabling root SSH access, creating a non-root user, configuring UFW firewall rules, and serving the Hermes agent dashboard securely over the Tailnet. Also explains direct vs. indirect prompt injection attacks and mitigation strategies like least-privilege API keys, separate accounts for agent actions, spending caps, and keeping secrets out of prompts. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How do I secure SSH access to a VPS running an AI agent using Tailscale? Install Tailscale on the VPS and any device that needs access, then edit the SSH config to set the listen address to the Tailscale IP, disable password authentication, and set PermitRootLogin to no. After restarting the SSH service, the server only accepts connections over the private Tailscale network, and attempts to SSH via the public IP time out. _Developers hardening AI agent infrastructure can find deeper security walkthroughs curated on daily.dev._ ### What is the difference between direct and indirect prompt injection in AI agents? Direct prompt injection happens when someone chats with an agent directly and tells it to ignore instructions or leak data, such as asking it to print its system prompt and API keys. Indirect prompt injection occurs when malicious hidden instructions are embedded in content the agent reads passively, like an email, tricking it into taking unauthorized actions such as forwarding messages. _Anyone building or hardening AI agents can track prompt injection defenses through daily.dev._ ### How can I limit the damage from prompt injection attacks on an AI agent? Give the agent least-privilege, read-only API keys wherever possible, use separate dedicated accounts (like a distinct Gmail) instead of personal ones, set spending caps on API keys, and keep secrets out of prompts by loading credentials into environment variables the agent never directly reads. Limiting what the agent can send outward minimizes the impact even if injection succeeds. _Teams weighing agent permission tradeoffs can follow ongoing AI agent security guidance on daily.dev._ ## Similar posts on daily.dev - [Agentic AI Security: Defending Against Prompt Injection and Tool Misuse](https://daily.dev/posts/agentic-ai-security-defending-against-prompt-injection-and-tool-misuse-8x1r1ytqn) · Machine Learning Mastery · 1 upvotes · 0 comments - [Set Up a Secure OpenClaw Deployment\!​](https://daily.dev/posts/set-up-a-secure-openclaw-deployment--xrwwrt3tm) · Daily Dose of Data Science \| Avi Chawla \| Substack · 2 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#ssh](https://daily.dev/tags/ssh), [#prompt-injection](https://daily.dev/tags/prompt-injection), [#tailscale](https://daily.dev/tags/tailscale) [View this post on daily.dev](https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f) ```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":"AI Agent FULL Security Guide - Don't Get Hacked","url":"https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f"},"datePublished":"2026-07-22T16:14:55.509Z","dateModified":"2026-09-14T06:52:36.415Z","description":"A comprehensive security guide for AI agent deployments covering three main attack surfaces: the server (VPS), prompt injection, and tool permissions. Covers...","image":"https://i.ytimg.com/vi/MEfpHENH78c/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/MEfpHENH78c/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Tech With Tim","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":"Tech With Tim","logo":"https://media.daily.dev/image/upload/s--6vVIDfU0--/f_auto/v1711726146/logos/TechWithTim","url":"https://daily.dev/sources/TechWithTim"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,ssh,prompt-injection,tailscale","timeRequired":"PT38M","video":{"@type":"VideoObject","name":"AI Agent FULL Security Guide - Don't Get Hacked","description":"A comprehensive security guide for AI agent deployments covering three main attack surfaces: the server (VPS), prompt injection, and tool permissions. Covers...","thumbnailUrl":"https://i.ytimg.com/vi/MEfpHENH78c/sddefault.jpg","uploadDate":"2026-07-22T16:14:55.509Z","duration":"PT38M","url":"https://api.daily.dev/r/0i3TOxn7F","embedUrl":"https://www.youtube.com/embed/MEfpHENH78c"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Tech With Tim","item":"https://daily.dev/sources/TechWithTim"},{"@type":"ListItem","position":3,"name":"AI Agent FULL Security Guide - Don't Get Hacked"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/ai-agent-full-security-guide---don-t-get-hacked-0i3toxn7f#faq","mainEntity":[{"@type":"Question","name":"How do I secure SSH access to a VPS running an AI agent using Tailscale?","acceptedAnswer":{"@type":"Answer","text":"Install Tailscale on the VPS and any device that needs access, then edit the SSH config to set the listen address to the Tailscale IP, disable password authentication, and set PermitRootLogin to no. After restarting the SSH service, the server only accepts connections over the private Tailscale network, and attempts to SSH via the public IP time out. Developers hardening AI agent infrastructure can find deeper security walkthroughs curated on daily.dev."}},{"@type":"Question","name":"What is the difference between direct and indirect prompt injection in AI agents?","acceptedAnswer":{"@type":"Answer","text":"Direct prompt injection happens when someone chats with an agent directly and tells it to ignore instructions or leak data, such as asking it to print its system prompt and API keys. Indirect prompt injection occurs when malicious hidden instructions are embedded in content the agent reads passively, like an email, tricking it into taking unauthorized actions such as forwarding messages. Anyone building or hardening AI agents can track prompt injection defenses through daily.dev."}},{"@type":"Question","name":"How can I limit the damage from prompt injection attacks on an AI agent?","acceptedAnswer":{"@type":"Answer","text":"Give the agent least-privilege, read-only API keys wherever possible, use separate dedicated accounts (like a distinct Gmail) instead of personal ones, set spending caps on API keys, and keep secrets out of prompts by loading credentials into environment variables the agent never directly reads. Limiting what the agent can send outward minimizes the impact even if injection succeeds. Teams weighing agent permission tradeoffs can follow ongoing AI agent security guidance on daily.dev."}}]} ```