yogsoth-ai/de-anthropocentric-research-engine

conceptual-blending

Construct multiple input spaces, extract a shared generic space, selectively project structure into a blend, and simulate emergent properties that are not present in either source alone.

ソースを見る

使い方

  1. インストールコマンドをコピーし、自分の環境へ導入する前にソースを確認します。
  2. 必要なツール、認証情報、対応環境を原文で確認します。
  3. 入力と成果物を指定し、出力を資料と照合します。
npx skills add https://github.com/yogsoth-ai/de-anthropocentric-research-engine --skill conceptual-blending

開始前の確認

  • 導入にはモデル契約、API枠、ファイルへのアクセス許可は含まれません。
  • スター数や導入数は活動の指標であり、独立した品質・安全性試験ではありません。
リポジトリの原文

見出し、例、コード、表、リンク、参照画像を含む原文を表示しています。

conceptual-blending

Purpose

Blend multiple input spaces through a generic space and selective projection to generate emergent research ideas.

Input contract

yaml
mode_contracts:
  two-space-blend: &blending_input
    required: [input_spaces]
    optional: [compatibility_constraints]
    constraints: [each_input_space_must_have_explicit_entities_and_relations]
  multi-space-blend: *blending_input
  emergent-property-search: *blending_input

Execution protocol

Do not perform called SOP operations inline; each loaded SOP owns its contract and thresholds.

  1. You MUST load skill construct-input-spaces to construct the input spaces.
  2. You MUST load skill extract-generic-space to extract the generic space.
  3. You MUST load skill simulate-emergent-properties to simulate emergent properties.
  4. You MUST load skill evaluate-compatibility to check compatibility.
  5. You MUST load skill synthesize-idea to synthesize the blend.

If the blend should be explored across a systematic combination space, consider explore-dimensional-space. If its claimed structural transfer needs formal testing, consider audit-structural-equivalence. Deviation: two-space and multi-space modes change only the number of inputs; generic-space extraction is mandatory.

Mode branches

  • two-space-blend: blend two explicit input spaces.
  • multi-space-blend: blend three or more spaces while preserving each projection trace.
  • emergent-property-search: prioritize properties absent from every source space alone.

Output contract

yaml
mode_contracts:
  two-space-blend: &space_blend_output
    produces: [input_space_set, generic_space, blend_candidates, idea_set]
    delta_fields: [findings, hypothesis_updates, uncertainties, decisions, open_questions]
  multi-space-blend: *space_blend_output
  emergent-property-search:
    produces: [input_space_set, blend_candidates, emergent_property_report, idea_set]
    delta_fields: [findings, hypothesis_updates, uncertainties, decisions, open_questions]

Thresholds and quality gates

  • B: all source spaces and projected relations are recorded; emergent properties must be absent from each source alone and compatible with target constraints.

Failure and counterexamples

Reject mere juxtaposition, unsupported emergence claims, and blends with unresolved relation conflicts.

Provenance map

  • creative-ideation/combinatorial-creativity, conceptual-blending, emergent-property-hunting: resolved/concept per exact v3 lookup.
  • Status: creative-ideation/combinatorial-creativity, emergent-property-hunting resolved; conceptual-blending concept (no exact v3 node).

Preserved source criteria ledger

  • Preserve generic-space construction, selective projection, and emergent-property semantics.

Context checkpoint / Delta notes

Append source-space changes, projected relations, emergence claims, compatibility findings, and selected ideas.

タスクに関連する項目

同じリポジトリから

関連する Skills

すべての Skills