{"id":830807,"date":"2026-11-11T04:32:12","date_gmt":"2026-11-11T04:32:12","guid":{"rendered":"https:\/\/www.newsbeep.com\/uk\/830807\/"},"modified":"2026-11-11T04:32:12","modified_gmt":"2026-11-11T04:32:12","slug":"14-year-old-creates-tool-for-ai-agents-to-band-together-like-a-team","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/uk\/830807\/","title":{"rendered":"14-year-old creates tool for AI agents to band together like a team"},"content":{"rendered":"<p>                    <a data-tag=\"ai and ml\" href=\"https:\/\/www.theregister.com\/tag\/ai%20and%20ml\" class=\"lab-article-section-link\" rel=\"nofollow noopener\" target=\"_blank\"><\/p>\n<p style=\"\" class=\" lab-article-section lab-article-section-top \">\n                    ai and ml\n                <\/p>\n<p>                    <\/a><\/p>\n<p class=\"subtitle below\" style=\"\">\n        Autark can act without constant prompting, but leaves users to enforce their own security boundaries\n    <\/p>\n<p>Built by a 14-year-old developer, a new tool called <a href=\"https:\/\/openautark.com\/\" rel=\"nofollow noopener\" target=\"_blank\">Autark<\/a> treats AI agents as a virtual team that can work without constant prompting.<\/p>\n<p>Users put tasks in an inbox, where a manager agent delegates them to other agents that carry out the work and report their results. Agents can also ask clarifying questions if tasks are too vague.<\/p>\n<p>\u201cAgents do not wait for a prompt,\u201d Autark\u2019s website <a href=\"https:\/\/openautark.com\/faq\/\" rel=\"nofollow noopener\" target=\"_blank\">explains<\/a>. \u201cA schedule, a timer, a new task, a message or an event wakes them up. They decide what to do, do it, and can plan their own next step.\u201d\u00a0<\/p>\n<p>As one would hope from a product like this, Autark is model agnostic and can be set up to connect to OpenAI Codex, Claude Code, Gemini CLI, GitHub Copilot CLI, and Kimi Code through those services\u2019 CLI tools. Local models are also supported. Your mileage may vary if you choose to use anything besides Codex or Claude Code, though, as Autark\u2019s website mentions support for other models is built in but hasn\u2019t been tested in a live run yet, as Autark is still under active development as an open-source tool.\u00a0<\/p>\n<p>Real teenage engineering<\/p>\n<p>Autark is the brainchild of teenage Austrian developer Stanislaus Hees. He\u2019s still in school, and tells The Register he\u2019s deeply interested in AI and came up with Autark after being unhappy with existing AI tools.\u00a0<\/p>\n<p>\u201cI could get help with individual tasks, but I still had to organise the work, remember what was unfinished and keep pushing it forward,\u201d Hees said of his time using Claude Code and Codex. \u201cI wanted a system that could take responsibility for ongoing work, not just respond to my next message.\u201d<\/p>\n<p>He developed Autark &#8211; or, more accurately, prompted AI coding tools to do it for him &#8211; to come up with a tool that could fulfill that want. You know what that means: It\u2019s capable, but that capability also makes it dangerous.\u00a0<\/p>\n<p>Autark\u2019s page on access and data makes clear what\u2019s at risk if you give it unfettered access to a PC, and why you don\u2019t want to let the thing run roughshod over a system.<\/p>\n<p>\u201cIt can do everything,\u201d the page <a href=\"https:\/\/openautark.com\/access\/\" rel=\"nofollow noopener\" target=\"_blank\">explains<\/a>, because \u201cAutark does not put walls around its agents.\u201d<\/p>\n<p>There\u2019s no sandboxing in Autark, no approval prompts, and no permission gates. It doesn\u2019t stop agents from spending all the money they can get their hands on, deleting files, or sending unsolicited emails, either. Agents run with the privileges of the machine&#8217;s user, with approval checks and permission restrictions disabled or bypassed. Even the safety limit parley that users have with Autark establishes guidelines, not rules.\u00a0<\/p>\n<p>\u201cThe first thing the AI asks is how far it may go,\u201d the access page points out. \u201cThat is a question, not a technical barrier.\u201d<\/p>\n<p>That means agents given a task that conflicts with their stated limits may still attempt to carry it out. As <a href=\"https:\/\/www.theregister.com\/security\/2026\/10\/02\/openai-alerts-100-orgs-that-its-misaligned-models-attempted-to-break-in-or-worse\/5300891\" rel=\"nofollow noopener\" target=\"_blank\">we\u2019ve seen in recent months<\/a>, some AI agents have found ways around restrictions, instructions be damned.<\/p>\n<p>\u201cIn our own live test runs, agents given a vague request reached real repositories, backups, SSH keys and .env files within minutes, and one changed the permissions of a file it had found,\u201d the access page notes.\u00a0<\/p>\n<p>In other words, it\u2019s probably best to stick Autark on a machine that doesn\u2019t contain anything you\u2019d want to avoid losing &#8211; even Hees admits as much.\u00a0<\/p>\n<p>\u201cI would not claim that \u2026 the app is suitable for unattended access to sensitive systems,\u201d the teen dev told us.\u00a0<\/p>\n<p>That\u2019s not to say Autark lacks protections for secrets, though. Autark\u2019s site explains that stored passwords and keys are encrypted and referred to using IDs that keep their values out of prompts. It also says its servers don&#8217;t store user data, with a meeting server used to relay encrypted traffic between devices. <\/p>\n<p>If you&#8217;re willing to try Autark to see whether it can improve your workflows, you don&#8217;t need to rely on its servers. While prepackaged apps are available from Autark\u2019s website, it\u2019s also fully open source and available <a href=\"https:\/\/github.com\/teydra\/OpenAutark\" rel=\"nofollow noopener\" target=\"_blank\">on GitHub<\/a>. Everything, even the meeting server code, is available for self-hosting and modification under an Apache open source license.\u00a0<\/p>\n<p>The bottom line is that this is a tool built by an ambitious young person, and it\u2019s a tool with a lot of potential if used correctly. That correct usage is the major challenge, though, so be sure to build your own guardrails.\u00a0 \u00ae<\/p>\n","protected":false},"excerpt":{"rendered":"ai and ml Autark can act without constant prompting, but leaves users to enforce their own security boundaries&hellip;\n","protected":false},"author":2,"featured_media":830808,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-830807","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/830807","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=830807"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/830807\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/830808"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=830807"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=830807"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=830807"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}