Use when user wants code reviewed for correctness, style, bugs, and maintainability. Triggers on: review code, code review, check PR, 審查程式碼, 檢查程式碼. Produces severity-classified findings with a verdict. Do NOT use for security-focused audit (prefer security-audit) or SQL-focused review (prefer sql-review).
Use when user reports a bug, error, exception, or unexpected behavior needing root cause analysis and minimal fix. Triggers on: debug this, why does this fail, root-cause this, 除錯, 找 bug. Performs systematic isolation and minimal fix. Do NOT use for feature requests (prefer implement) or known simple typos (prefer…
Use when user needs code written — new features or task execution in Java 8 / Maven / Spring / Hibernate projects. Triggers on: implement, code this, write code, 實作, 幫我寫. Produces working code following existing patterns. Do NOT use for refactoring without new behavior (prefer refactor) or bug investigation (prefer…
Use when user needs a phased implementation plan with requirements, file impact, risks, and alternatives before coding — including clarifying vague or ambiguous requirements first. Triggers on: plan, how should we build, unclear requirements, 規劃, 幫我想方案. Produces a structured plan document. Do NOT use for atomic task…
Use when existing code needs structural improvement without changing behavior — extract method, rename, eliminate duplication, simplify logic. Triggers on: refactor, clean up, extract method, 重構, 整理程式碼. Produces behavior-preserving structural changes. Do NOT use for new features (prefer implement) or bug fixes (prefer…
Use when user needs an OWASP-focused security audit of Java web code — injection, auth, access control, and configuration review. Triggers on: security audit, OWASP, vulnerability check, 資安審查, 有沒有漏洞. Produces severity-classified security findings. Do NOT use for general code review (prefer code-review) or SQL-only…
Use when user needs SQL reviewed — queries for injection risks, performance, and index strategy, or DDL/DML migration scripts for rollback safety, data-loss risk, lock impact, and backward compatibility. Triggers on: review SQL, slow query, review migration, SQL 審查, 看 migration. Produces severity-classified findings…
Use when a plan needs to be broken into dependency-ordered atomic tasks for implementation. Triggers on: break down tasks, task list, decompose, 拆任務, 任務拆解. Produces a T###-formatted task list with dependency ordering and parallel markers. Do NOT use for creating plans (prefer plan) or direct implementation (prefer…
Use when checking whether an implementation actually meets its acceptance criteria — derive the checks, bind each to a runnable command, execute, and gate pass/fail so a build loop has an objective exit condition. Triggers on: verify, does it pass, acceptance check, definition of done, test cases, what to test, 驗證…