Task Verification by Acceptance Criteria

★ 7.7 · testing

verify-task runs code verification in Claude Code for a specific task defined in EXECUTION_PLAN.md, following a structured eight-step verification workflow. It accepts a task ID, extracts all acceptance criteria, and optionally reads configuration from .claude/verification-config.json to determine available test, lint, type-check, and build commands. The workflow covers TDD compliance (including git-history test-first checks), an optional mutation-testing quality gate, criterion-by-criterion verification across CODE, TEST, LINT, TYPE, BUILD, SECURITY, BROWSER, and MANUAL types, and finally logs results to verification-log.jsonl. For browser-based criteria, it checks tool availability across ExecuteAutomation Playwright, Browser MCP, Microsoft Playwright, and Chrome DevTools, offering a graceful fallback to manual verification. The skill is designed for teams that manage work through execution plans and need to verify individual task acceptance criteria inline, without spawning sub-agents.