| name | creative-synthesis |
| description | ๐ก Creative synthesis across domains. |
Creative Synthesis
Open the possibility space without losing the purpose.
Begin from the desired change, materials, constraints, existing candidates, and work already attempted. Form a provisional creative brief and ask only for an unknown that changes what novelty must do.
Read references/creative-synthesis-doctrine.md when the field is stuck, analogies carry the reasoning, constraints are contested, or candidates are collapsing into cosmetic variants. Use assets/concept-field.template.md when several candidates, people, revisions, or experiments must remain legible.
Diverge across different generative moves: reframe the purpose, invert an assumption, transfer relational structure from another domain, recombine functions, change scale or time, shift actor or viewpoint, exploit an exception, or let a hard constraint create form. Make each leap explicit enough to inspect. Prefer a small field of orthogonal concepts over a long list of renamed defaults.
Hold convergence until the field has real difference. Then develop the strongest concepts through purpose fit, mechanism, tensions, assumptions, failure edges, and the cheapest experiment that would teach something. Preserve a strange but fertile candidate when it opens a distinct region, even if it is not immediately practical.
Treat analogies as models, not mechanisms: map relations, test breaking points, and name smuggled entailments. Treat candidates as proposals, not facts or commitments. Route warrant to Epistemic Regulation and selection to Decision Intelligence. Use Agent Dreaming only when the user explicitly authorizes its packet-closed, higher-variance offline mode; ordinary task-time invention stays here.
Finish when the concept field is differentiated, purpose-tethered, and ready for evaluation or choice, with its generative logic and next experiments visible.