okhp3-refolddec-core

okhp3-refolddec-core

OverKill Hill P³ · overkillhill.com · github.com/OKHP3

Transformation-aware execution for fold, unfold, and refold operations across representation types.

1. Identify the Transformation

Before executing, name the operation clearly:

Fold — the source is a complex artifact; the target is a more abstract or compact form. Examples: process narrative → SKILL.md, set of observations → taxonomy, long document → diagram.

Unfold — the source is a compact or abstract artifact; the target exposes its constituent parts. Examples: SKILL.md → step-by-step procedure, diagram → process narrative, taxonomy → flat item list.

Refold — the source and target are in different representation types; semantic content is preserved across the change. Examples: diagram → documentation, documentation → SKILL.md, SKILL.md → structured prompt, process → diagram.

State the transformation type explicitly before proceeding. "I am refolding [source type] into [target type]."

2. Inventory the Source

Before transforming, inventory what the source artifact contains:

• Entities — what things are named or described
• Relationships — how those things connect, depend on, or act on each other
• Conditions — decision points, exceptions, and branching logic
• Constraints — rules, limits, and boundaries that must be preserved
• Audience-specific content — material that is calibrated for a specific reader and must be recalibrated for the target form

This inventory is the fidelity baseline. Semantic loss is measured against it.

3. Track Semantic Loss

Every transformation loses something. Name the losses before committing.

For each inventory item, assess: does the target representation support this? If not, it is a loss.

Low-loss transformations: The target representation can hold all inventory items with no change in meaning.