Detective Bots Simulation.

A conversational multi-agent experiment exploring personality prompts, group dialogue, and voting in a detective game.

Explore source on GitHub Read the README

Bar chart of detective-game outcomes across the repository's documented model runs — Game-outcome figure from the repository's bounded 40-run experiment.View original on GitHub

ENGINEERING DEEP DIVE

Inside the system.

01 / THE CONTEXT

A bounded detective game provides a way to study how personality prompts affect language-model dialogue and group decisions.

02 / THE APPROACH

Personality-conditioned agents share a dialogue history under a game controller. The controller rotates questions, collects votes, and checks outcomes. Saved JSON and CSV records support game-outcome plots and topic analysis across 40 documented runs.

03 / EXPLORE FURTHER

Explore the controller, persona prompts, saved dialogue, and outcome analysis. The plotted results describe the repository's 40-run experiment and its specific model and prompt setup.

FROM THE REPOSITORY

What's inside.

01
Orchestrates personality-conditioned agents through a shared game controller.
02
Saves dialogue and outcome records for analysis across 40 documented game runs.
03
Includes model-outcome plots and topic-analysis code for this bounded experimental setup.

These notes summarize the reviewed implementation and available artifacts. Open the original source for code, documentation, and subsequent changes.

Open the original repository

Inside the system.

What's inside.

Let's make it work.