Clawdbot documentation expert with decision tree navigation, search scripts, doc fetching, version tracking, and config snippets for all Clawdbot features
Executes tasks in continuous loop mode with strict test coverage and no legacy compatibility. Use when the user says "进入loop 模式, 目标: XXXX" or asks for nonstop implementation until done.
Red-team review of system/design proposals to find fatal flaws and blockers across architecture/tech risk, business reasonableness, and security/compliance. Outputs rejection-style critique (blockers, contradictions, missing invariants), not improvement plans. Use when the user wants a harsh risk审查/否决/质疑/拆穿 review.
Review business goals, PRDs, business flows, user journeys, and UX/interaction designs for completeness, clarity, logical consistency, and alignment with user needs. Use when the user asks to review a business goal/design proposal, PRD, process flow, user journey, or interaction plan.
Reviews designs and business goals for security vulnerabilities, data protection (in transit/at rest), authorization, and compliance alignment. Use when the user asks for a security review, threat modeling, attack surface analysis, data leakage prevention, or compliance/security assessment.