What problem does it solve?
This Skill eliminates the critical "code 100% ≠ live 100%" trap where applications pass all static checks, type checks, and test suites yet still fail in production due to hydration mismatches, missing interaction validations, and surface-level browser testing. It prevents false confidence by enforcing deep, multi-round live verification with Playwright MCP.
Core Features & Use Cases
- 9 Inspection Modes: Comprehensive coverage including UI/UX, database, code quality, API, accessibility, performance, SEO, security, and integration testing.
- Autonomous N-Round Engine: Runs 5-stage verification loops until all 4 layers (health, Playwright interaction, DB schema, code quality) pass, with Mode A for manual review and Mode B for autonomous ScheduleWakeup cycles.
- Meta-Learning Trap Prevention: Permanently cites R45-R77 meta-learning patterns at the start and end of every call to prevent recurring verification mistakes like curl-only false positives and self-defined layer bypass.
- Persona Auto-Bootstrap: Automatically maps and loads domain-specific personas with 3-step approval for missing personas.
Quick Start
Use the live-verify-loop skill to start an autonomous verification audit by typing '/live-verify' or saying 'run live verification loop', then follow the interactive 7-step settling to configure modes and personas before the N-round engine begins.