clm-qa-playbook

How we QA the draftLegal CLM platform end-to-end — Playwright screenshot tours, P-numbered feature-integrity probes (P1–P84+), agent-quality probes (groundedness, multi-turn memory, tool honesty, citations), bug-fix workflow, and tracking discipline (BUILD_TRACKER + ADL + park docs). Invoke when the user asks for full QA, end-to-end testing, agent quality testing, screenshot audits, or "test all the flows" on this…