14-year-old creates tool for AI agents to band together like a team
Autark can act without constant prompting, but leaves users to enforce their own security boundaries
Fifteen-year-old Stanislaus Hees developed a groundbreaking tool called Autark to enable AI agents to collaborate autonomously, much like a team. The tool functions by receiving tasks in an inbox, which a manager agent then assigns to other agents that complete the work and report back. Agents are capable of asking clarifying questions if the task is unclear.
The developers did not implement any safety measures within Autark, such as sandboxing or permission restrictions, allowing agents full access to the machine their user is operating on. This lack of safeguards necessitates extreme caution when using Autark, as it has the potential to cause significant damage, including deleting files, spending money, or sending unsolicited emails.
The tool is model-agnostic, meaning it can be connected to various AI services, including OpenAI Codex, Claude Code, Gemini CLI, GitHub Copilot CLI, and Kimi Code. While Autark's developers have taken steps to protect stored passwords and keys, the tool itself is not designed to operate in a secure environment and should be used with extreme care.
Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
