What problem does it solve? When a single agent runs an analysis or generation skill single-threaded, every evaluation dimension only gets shallow coverage because context is limited. This Skill deepens execution by splitting the target skill's work into independent dimensions, dispatching sub-agents to investigate each in depth, then having the main agent cross-verify and merge the findings. ## Core Features & Use Cases - Output Amplification (analysis skills): Splits evaluation dimensions of skills like code-review or design-review, dispatches read-only sub-agents per dimension, then runs coverage checks, conflict detection, and evidence verification before producing the final report. - Input Amplification (generation skills): Parallelizes only the research phase of writing-oriented skills like design-craft, collecting evidence packages while the main agent still writes single-threaded to preserve global consistency. - Respect for Native Discipline: Honors any parallel/serial rules already built into the target skill, keeps serial-only dimensions on the main agent, and always outputs in the target skill's native report format. - Use Case: A user says "use amplified mode to review this commit" — the Skill locks onto code-review, keeps its serial dimensions on the main agent, dispatches five sub-agents for security, bug-risk, style, performance, and testing, verifies high-severity findings against raw evidence, and outputs a standard Review Report with an amplification note. ## Quick Start Ask the agent to run a specific skill in amplified mode, for example by saying "use amplified mode to run code-review on this change".
