Build Log · June 25, 2026

The Principle Engine — vocabulary before the machine

Took the first disciplined step into the principle-discovery engine the last session flagged as 'the next game' — by defining the words a claim moves through (habit, pattern, primitive, principle, law) instead of building the state machine that moves them.

  • ForgeKit OS
  • Transfer Research
  • Principle Engine
  • Strike
new for-zeb file (vocabulary doc)
1

new for-zeb file (vocabulary doc)

terms defined by kind
5

terms defined by kind

Strike + Anvil captured
1

Strike + Anvil captured

cheap kit edits applied
2

cheap kit edits applied

Timeline

  1. Start

    Read the whole instrument

    Explorer kit (Tim-facing + for-zeb), focus.md, the three prior 06-25 retros, anvils + strikes

  2. Diagnosis

    Found the mechanism written 3x, partially

    graduation test (research-track), promotion/demotion (evidence-framework), Promotion (THE-LOOP) — none owns it

  3. Decision

    Two choices confirmed with Zeb

    Define vocabulary first (not the state machine); product-impact gate is a hard requirement for 'law'

  4. Build

    One vocabulary doc + 2 cheap edits

    Five terms by kind; deferred demotion + contradiction-handling to the engine session

  5. Harvest

    Anvil + Strike to the journal

    The engine is a transition that already exists unnamed — define the words before the machine

What shipped

  • New for-zeb/principle-engine-vocabulary.md — defines habit, pattern, primitive, principle, law by what they ARE (in kind, not just confidence), with the chain shape sketched for reference and the state machine explicitly NOT built

  • Answered Zeb's nine principle-engine questions: 5 defined here, demotion + contradiction-handling deferred to the engine session (run against a real claim from Tim, not an imagined one)

  • Locked the product-impact gate: no claim reaches 'law' until it has demonstrably changed how a shipping app (Forge, Leashline, ForgeHome) got built — the structural defense against a research org that produces research instead of products

  • research-track.md headline reframed from 'How much of ForgeKit is transferable?' to 'What survives across different builders?' — ForgeKit is the instrument, not the subject

  • Added a balancing rule to research-track.md: every quarter ask 'what did the research make POSSIBLE?' not 'what did we discover?' — keeps the lab attached to the product spine

  • Anvil + Strike captured: the Principle Engine is a transition that already exists unnamed, not a new layer; define vocabulary before building the machine

“The Principle Engine is a transition that already exists unnamed, not a new layer to add — and you define the vocabulary before you build the machine that moves claims through it.”
— The Strike this session captured