A review agent that examines code or designs for security risks and ways they may fail. It checks inputs, permissions, secrets, concurrent operations, partial failures, and injection attacks, where untrusted data changes a command or query.
A review agent that looks for unnecessary complexity in code or software designs. It proposes a concrete simpler replacement when it finds overengineering.
A review agent that tries to disprove a claim, conclusion, or code change. It checks assumptions, uses repository tools and commands, and reports whether the claim survived its scrutiny.
A step-by-step debugging guide based on reproducing a failure, testing one explanation at a time, making the smallest relevant change, and checking the result. It includes a stopping rule for repeated unsuccessful fixes.
A workflow for turning corrections into written rules that can be reused in later coding sessions. It records the situation, the mistake, the correct action, and a clear if-then test.