v9.3.0: V10 Agentic Reform

Changed

• AGENTS.md: full rewrite — caveman-style throughout, Main Directive at top, Golden Rules tightened, feature lifecycle phases documented, runtime artifact resolution convention, review loop relay protocol, loop exit detection in anchor, multi-skill ordering explicit, beehave violations = test failures, convention boundary for design-phase states
• refactor/SKILL.md: added beehave check before task test-fast (step 9), added loop exit detection via beehave status --json (step 10)
• write-test/SKILL.md: added no-docstrings rule for test functions with self-heal ruff instruction
• review-gate/SKILL.md: added no-docstrings enforcement in structure review (step 8)
• structure-feature/SKILL.md: clarified out: git_branch as Runtime artifact meaning branch creation
• todo-anchor-protocol.md: synced notation to native status fields, added loop resolution and review relay to anchor checklist
• test-design.md: added comment/docstring noise pattern to Spec Value Fidelity
• test-stubs.md: documented auto-generation on every pytest, stale stub accumulation, beehave clean --force usage
• skill-design/principles.md: added imperative language standard for skill writing
• pyproject.toml: added "D" to per-file-ignores for tests/** (no docstrings required in tests)

Simulation-Driven

• Walked full 24-state define-flow → develop-flow path for "Add Task" feature
• Identified and resolved 5 critical pain points: runtime artifact ambiguity, feature file lifecycle, multi-skill ordering, review feedback survival, loop exit detection