Governance-Driven Design

🔒

Governed Constraint Set

The primary artifact — what the system must hold true, what it must not violate, and what remains unresolved. Produced before any agent runs, referenced by everything downstream.

🔄

Iterative Constraint Refinement

A six-step cycle that formalizes domain assumptions, stress-tests each one, checks internal consistency, and surfaces what cannot be resolved without human judgment.

🎯

Unresolvable Residue

The most valuable output of a GDD cycle is what it cannot resolve. Surfacing it before build is the discipline. Naming its owner is the governance act.

🧠

Human-as-Primitive

The Synthesis Gate is not an interrupt. It is a designed-in architectural node — the moment the machine hands the unresolvable residue to a human who can resolve it.

📋

CLAUDE.md Template

Drop one file into your project root. Make Claude Code governance-aware before the first ICR cycle runs. Step zero is separable from step one.

🔗

Upstream Completion

Not a replacement for TDD, BDD, or DDD. Their upstream completion — the discipline that stays ahead of AI-first delivery velocity.

The Governance Gap

AI-first delivery is fast. Dangerously fast. Agents generate artifacts at a pace that outstrips the organizational capacity to understand what was built, why, and whether it is correct.

The response has largely been reactive: add governance after the fact. This produces the governance gap — the distance between what an AI system was instructed to do and what it can be held accountable for doing.

The governance gap is not a tooling problem. It is a sequencing problem.

Governance applied after execution is remediation. Governance applied before execution is design.

The Lineage

TDD  (2003)  →  correctness defined at implementation
BDD  (2006)  →  correctness defined at behaviour
DDD  (2003)  →  correctness defined at domain model
GDD  (2026)  →  correctness defined at governance layer
               before any artifact exists

Each transition moved correctness earlier. GDD is the next step — the logical response to AI-first delivery compressing the gap between "designed" and "in production" to near-zero.

The Cycle at a Glance

1. FORMALIZE   write every assumption explicitly
2. STRESS      find what breaks each one
3. CHECK       verify internal consistency
4. SURFACE     collect what cannot be resolved
5. GATE        human decision on the residue
6. CONVERGE    exit with explicit acknowledgment

Exit condition: not all resolved — but all residue explicitly acknowledged with an owner.

Governance-Driven DesignFormalize intent before the machine runs.