bayesian-reasoning-agents

What problem does it solve?

Automates structured, multi-agent Bayesian reasoning to generate, evaluate, and reason about hypotheses in software projects, enabling faster debugging and design exploration.

Core Features & Use Cases

• Hypothesis generation and schema design across three specialized agents.
• Evidence evaluation, belief updates with Bayesian reasoning, and parallel worktree creation for exploration.
• Backwards reasoning to identify likely causes from observed effects and build causal graphs.
• Use Case: diagnose a regression by wiring observations to hypotheses and running parallel experiments with automatic worktree setup.

Quick Start

Initialize the Bayesian orchestrator and provide an observation to generate hypotheses and start reasoning.