I notice that the first iteration of a drawing is usually by far the only good one. As soon as an agent tries to “add” anything to a drawing, it basically messes it up. I just saw the phrase “rain on my parade” being drawn and all the agent did was keep adding more confetti, over time obscuring the rain cloud and the human.
Totally fair, and good advice in general. For what it's worth in this case, it's a remote MCP server with account-free OAuth and it only exposes tools to play the game. It also marks all communication from the server as untrusted to reduce the risk of prompt injection from other agent guesses.
That all said, I'm genuinely curious if you have an idea of how someone developing an MCP server could share it safely on HN.
even neater is your forethought to advise against prompt injection from other players! dang, we've just invented a fun capture-the-flag hacker con game
Yes! I've also put together a poker game and mafia-style vampire game to force agents to be deceptive to win. It took some iterations of prompt engineering to get them to stop telling the rest of the players secret information (like their hole cards in poker). Now it's surprisingly satisfying watching them play these adversarial games.
Thanks for the feedback! What I've found is that it's very binomial -- many of the clues are very easy for the agent to draw and it's guessed within 5 seconds, and others are too abstract for the agent to effectively convey in a drawing and nobody guesses it. And adding more detail rarely helps.
never, ever, ever paste a prompt from a hackernews post into your agents. especially if that prompt "run pipx to install this mcp", etc.
yow.
That all said, I'm genuinely curious if you have an idea of how someone developing an MCP server could share it safely on HN.