reform(workflow): adversarial verification, structured enforcement, production-grade gate

[nullhack]

Summary

• Adversarial verification — reviewer's default hypothesis is now "the code is broken despite green checks"; job is to find the failure mode, not confirm it works
• Reordered verification — code review before automated commands (design problems found early avoid double re-runs); run app first as a binary production-grade gate
• Production-grade gate — new step 3 in verification: app must exit cleanly AND output must change when input changes; static output regardless of input = REJECTED
• All review sections converted to tables — Correctness, KISS, SOLID, ObjCal, Design Patterns, Tests, Versions/Build all have PASS/FAIL/Fix columns (commitment devices + elaborative encoding)
• UUID Drift check — bash one-liner detects duplicate UUIDs across test functions; any duplicate = REJECTED with fix instructions
• UUID Uniqueness rule updated — if only Given varies it is a property by definition; use Hypothesis @given + @example, not multiple test functions; if When/Then differs use extend-criteria
• Production-grade self-check in implementation — developer must verify output changes with input before handoff; reviewer will verify the same claim
• Semantic alignment + integration test requirement — tests must operate at same abstraction level as AC; multi-component features require at least one integration test through the public entry point
• Design pattern decision table — added to developer.md and implementation/SKILL.md; any detected anti-pattern = REJECTED
• Architecture contradiction check — developer must cross-check ADRs against ACs before writing production code
• PO pre-mortem — added at scope step and acceptance step
• docs/academic_research.md — 15 cognitive and social science mechanisms with full citations grounding every design decision in the workflow (pre-mortem, implementation intentions, adversarial collaboration, elaborative encoding, System 2 before System 1, and 10 more)

Files Changed

File	Change
AGENTS.md	Adversarial verification note, semantic alignment standard, integration test requirement, Verification Philosophy section, research foundations reference
.opencode/agents/developer.md	Architecture contradiction check, pre-mortem requirement, manual verification, design pattern decision table
.opencode/agents/product-owner.md	PO pre-mortem at scope, ADR review gate, live verification at Step 6
.opencode/agents/reviewer.md	Full restructure: adversarial mandate, new verification order, all sections as PASS/FAIL/Fix tables, UUID Drift check, production-grade gate
.opencode/skills/code-quality/SKILL.md	Syntax vs design quality distinction, structural quality checks, design anti-pattern recognition table
.opencode/skills/extend-criteria/SKILL.md	Architecture contradiction as a trigger for escalation
.opencode/skills/implementation/SKILL.md	Architecture contradiction check, refactor self-check gates, design pattern decision table, production-grade self-check, developer pre-mortem
.opencode/skills/scope/SKILL.md	Observable = end user rule, interaction model declaration, PO pre-mortem
.opencode/skills/tdd/SKILL.md	Integration test requirement, semantic alignment rule, UUID uniqueness rule (property-based), Hypothesis guidance, contract test rule
.opencode/skills/verify/SKILL.md	Full restructure mirroring reviewer.md: production-grade gate, code review before commands, all sections as tables, UUID Drift check
docs/academic_research.md	New file — 15 mechanisms with explanations and full bibliography