cc10x v11.1.0 — execution-engine harvest + 4 net-new skills

Minor release — all changes are additive over v11.0.0. Driven by a system-level architecture comparison against superpowers + matt-pocock that honestly mapped where cc10x leads (back-half verification: independent verifier + REVERT + test-honesty gates, parallel silent-failure-hunter, fail-closed hooks) and where it trailed (the execution engine). This closes the engine gap without disturbing the per-phase build loop, which stays the default.

Execution engine

• New runnable scripts cc10x_phase_brief + cc10x_review_package (dependency-free) — produce the dispatch-by-reference brief and the recorded-BASE..HEAD diff package. Resolves a live contradiction where the policy + reviewer relied on a diff package no script produced.
• Recorded-BASE diffing across 4 agents — kills lossy HEAD~N / working-tree truncation that dropped earlier commits of a multi-commit TDD phase from review.
• Reviewer SPEC_COMPLIANCE verdict split (Missing / Extra / Misunderstood), separate from code quality.
• Inline no-subagent execution fallback — keeps the fail-closed gates when subagents are unavailable or work is too coupled to split.
• Per-TASK dispatch as an opt-in mode (per-phase stays default), optional final whole-branch review, on-BLOCKED-change-something rule, feedback-loop-first debugging, brainstorming glossary + synthesize-now.

Four net-new skills (19 → 23)

• receiving-code-review — verify human/external feedback before agreeing (closes the ungoverned PR-comment loop).
• codebase-deepening — deletion-test retrofit for shallow LLM-grown modules.
• frontend-design-critique — two isolated assessments + anchored rubric + AI-slop ban-list, RTL-aware.
• handoff-package — portable secrets-redacted cross-tool/session handoff.

Four validators green at every commit; live harness 4/4 PASS.

🤖 Generated with Claude Code