Got Controller

What it does

The Got Controller skill enables AI agents to manage and orchestrate complex research using a Graph of Thoughts (GoT) framework. It allows for strategic generation, aggregation, refinement, and scoring of research findings modeled as nodes in a graph, optimizing information quality through parallel exploration and iterative improvement. This facilitates managing multi-agent research workflows by controlling operations on the GoT structure.

When to use it

• Initial exploration of a topic, creating multiple research paths simultaneously.
• Combining related findings from different agents into a cohesive synthesis.
• Improving existing research findings through clarification and better organization.
• Managing graph complexity and focusing resources on high-quality research paths.

Key capabilities

• Generate(k): Creates k new research paths from a parent node, spawning parallel research agents.
• Aggregate(k): Combines k nodes into one stronger synthesis, resolving conflicts and extracting key insights.
• Refine(1): Improves an existing finding by enhancing clarity, completeness, citations, and structure.
• Score: Evaluates the quality of a research finding on a scale of 0-10 based on criteria like source quality and comprehensiveness.
• KeepBestN(n): Prunes low-quality nodes, retaining only the top n at each level to manage graph complexity.

Example prompts

• "Generate 4 research paths from this initial finding."
• "Aggregate these three nodes into a single synthesis."
• "Refine this node to improve its clarity and citations."
• "Score this research finding based on the provided sources."

Tips & gotchas

• The skill operates within the Graph of Thoughts (GoT) framework, which models reasoning as a graph. Understanding GoT concepts like nodes, edges, scores, and frontier is helpful for effective use.
• Scoring criteria are important; findings with multiple high-quality sources and no contradictions receive higher scores.