Relationships-First Prompting
The agent now prioritizes artifact relationships when reasoning, producing more connected and context-aware outputs.
The agent's reasoning model now starts from relationships between artifacts rather than treating each artifact in isolation.
What Changed
When the agent processes a request, it first maps the relationship graph around the relevant artifacts — what feeds into what, which templates produce which instances, what workflows depend on which data. This context-first approach produces more accurate, connected outputs.
Template Molecule Awareness
Templates now understand their full "molecule" — the cluster of related artifacts that form a complete unit. When you create from a template, the system pulls in the right connected pieces automatically instead of just cloning the template in isolation.
Fallback Tool Promotion
If the agent's primary approach doesn't have the right tools available, it now promotes fallback tools more intelligently based on the relationship graph, reducing dead-end responses.