AI agent skill · by @yogsoth-ai

Construct Input Spaces

Define the structured input spaces for conceptual blending: entities, relations, goals, constraints, and salient dynamics in each source.

Archived 2026-10-23 · Source updated 2026-09-29 · License: Apache-2.0. Check the original repository for the latest version.

SKILL.md instructions

# construct-input-spaces
## Purpose
Define the explicit source spaces used by conceptual blending.
## Input contract
```yaml
required: [source_domains, blend_goal]
optional: [entities, relations, constraints, salient_dynamics]
constraints: [each source needs entities, relations, goals, and constraints or an explicit unknown]
```
## Procedure
1. Partition each source domain into entities, relations, goals, constraints, and dynamics.
2. Normalize equivalent roles without erasing source-specific structure.
3. Record the blend goal and the items eligible for projection.

If the input spaces and their correspondences are explicit, consider `extract-generic-space` as the next tactic.

## Output contract
```yaml
produces: [input_space_set, normalized_roles, projection_candidates]
delta_fields: [findings, hypothesis_updates, uncertainties]
```
## Quality gates
- Every input space is independently reconstructible and source-specific relations remain visible.
## Failure and counterexamples
Reject an input space that is only a keyword list or that merges incompatible roles before blending.
## Provenance map
- `conceptual-blending/input-space-construction`: resolved.