• snoons@lemmy.ca
    link
    fedilink
    English
    arrow-up
    17
    arrow-down
    1
    ·
    18 hours ago

    I’ve seen one where it basically told the ai to halt and catch fire, so I usually check what’s inside just in case… mostly though, yeah.

    • Kangae_Hishiryo
      link
      fedilink
      arrow-up
      3
      ·
      4 hours ago

      Yeah. While there are adversarial AGENTS.md files, and you can recognize that an AGENTS.md is adversarial not only by reading its contents, but also by seeing if the project has a proper anti-AI police (so, basically, a totally zero-AI commitment, with no exceptions), they’re not that much.

      So yeah, it’s a good rule of thumb honestly.