Patterns / modal-constructs sibling surface — parked sketch 2026-05-27

DARE.CO.UK · PARKED SKETCH · 2026-05-30

Mirrored from ~/.claude/.../memory/parked_sketch_patterns_modal_constructs_surface_2026-05-27.md. This is a design sketch parked for future build — read for context, not as a current deliverable.

Dan named “patterns = the modal constructs” while scoping kb.gf.cx to raw memory only. The modal-constructs surface — extracting and presenting the higher-order cognitive shapes that emerge FROM the memory corpus (frames, schemas, ways-of-thinking) — is a future sibling subdomain to kb. Parked until the corpus is dense enough and the right cuts become obvious.

The frame

Dan, 2026-05-27, after I proposed adding pattern toggles to kb.gf.cx:

“Maybe it best to keep kb purely focused on memory/claude. And patterns is the modal constructs”

Two surfaces, two purposes:

Why this is parked, not killed

• The memory corpus is still actively growing — premature crystallization would lock in shapes that aren’t yet load-bearing.
• The right “cuts” (which patterns deserve their own modal-construct entries) emerge from operation, not analysis at rest.
• Dan operates by finding edges (per user_sketch_early_find_edges_scale.md) — the patterns surface will materialize when the edges are visible enough to draw.

What this surface could look like (sketch only)

• Domain candidate: patterns.gf.cx or constructs.gf.cx (modal-constructs is a mouthful)
• Each “modal construct” page documents one mental model: name, when-it-fires, the 3-5 memory files that ground it, examples
• A pattern entry isn’t a memory file — it’s a synthesis OVER multiple memory files
• The list of constructs IS the editorial output. Examples Dan already operates by:
• “Bake at build, ask at runtime”
• “Static-site + living brains”
• “Sketch / principle / toolkit”
• “Audit before batch”
• “Park rather than kill”
• “Cost-flattening via primitive composition”

Resume conditions

Build this when ANY of: 1. The same 3-4 memory files keep needing to be cross-linked from a new memory — that crowd-pulling is a signal a modal-construct page would consolidate them 2. Dan asks for a “thinking-models map” or “what frames am I operating by” view 3. A third party (collaborator, future-Dan, employee) needs to read “how Dan thinks” without wading through 460 files 4. The kb.gf.cx filter starts feeling inadequate because users are searching for cognitive frames, not artifacts

Cross-refs

• user_kb_gfcx_as_home_for_shared_memory.md — kb hosts artifacts
• feedback_kb_gfcx_scope_memory_only_no_synthesis.md — kb does NOT host modal constructs
• user_knowledge_to_wisdom_via_distillation.md — wisdom is what modal-constructs ARE; the surface would expose the distillation
• user_memorializing_thinking_as_2026_asset.md — modal-constructs surface is the 2026 differentiator made navigable
• feedback_sketch_principle_toolkit_pattern.md — itself a modal construct; would be a flagship entry