What problem does it solve? AI agents often misjudge how much rigor a task needs, either over-engineering trivial edits or under-verifying high-risk changes. This Skill provides a decision contract that matches execution depth, skill selection, reference usage, and independent review to the actual risk and scope of the work. ## Core Features & Use Cases - Execution Mode Policy: Classifies work into FAST, NORMAL, or DEEP modes with explicit verification budgets for each. - Risk-Gated Review: Applies RISK_0 through RISK_4 classifications and triggers independent audits for high-consequence actions like auth, payments, or production changes. - Just-in-Time Skill Selection: Selects only justified skills, references, and tools without keyword classifiers or fixed quotas, with an economy reviewer reference for low-cost decision audits. - Use Case: Before implementing a payment flow change, the agent records a compact task decision, loads the security gate skill, and requests an independent review before execution. ## Quick Start Ask the agent to record a task decision with execution mode and risk level before starting any material code change.
