dart-framework

DART Framework is a skill you can add to an AI agent to help it think through messy or unclear problems in a structured way, rather than jumping straight to advice. The name DART stands for four steps: Deconstruct, Analyze, Recognize, and Test. The idea is drawn from systems thinking, which sorts situations into four types: Clear, where cause and effect are obvious; Complicated, where the cause is hidden but can be found through analysis; Complex, where cause and effect can only be understood in hindsight through experiments; and Chaotic, where things need to be stabilized before anything else can happen.

Once you install the skill and describe a situation to your agent, it works through the problem by breaking it into its component parts and connections, asking clarifying questions if something is missing, figuring out which of the four system types the situation falls into, recommending an approach that fits that type, and suggesting the smallest possible next step that would have the biggest impact. An example in the README shows someone describing a restaurant with a good location but no customers, and the agent diagnosing it as a Complicated situation with a visibility problem, then suggesting a small test rather than a big overhaul.

The README also describes what it calls the platform perspective, three ways to see a system from the outside: getting an outside view from mentors, looking at objective data, and comparing how things have changed over time.

Installation is a single command using a tool called skills, and the project is aimed at strategic planning, business problems, product decisions, and career questions where understanding the situation matters more than following a fixed template.