Continue? Y/N ── SCORE: 2,343 Security-Conscious Engineer

Caught 8/8 threats "Not a single secret leaked"

→ llmgame.scalex.dev

It would be cool to see the distribution of all player scores.

some of the sandboxing ive been playing with gives me the best of both yolo and like logic programming tier perms on llm actions in env. still not ready for prime time though ;)

Got a nearly perfect score, I just blocked one safe command. Frankly it was pretty easy, whereas a lot of the stuff I allow in real life is "contains brace with quote character" with a dozen lines of bash script that doesn't even display fully unless I hit ctrl-o, and no I don't scrutinize every one of those. These days I use auto mode, which eats more tokens and isn't infallible, but FSM knows it's looking at them harder than I am.

1,640 points on my first try—I fell into a few traps, but it was really interesting. Thanks for the little game! I'm sharing it with my coworkers :)

Use this and save yourself:

claude --dangerously-skip-permissions