What problem does it solve? Coordinating multiple AI subagents on a shared codebase causes parallel build collisions, conflicting file edits, and premature commits. This Skill establishes strict orchestration rules so subagents write code safely while the orchestrator owns builds, tests, and commits. ## Core Features & Use Cases - Mandatory Subagent Use plan section: Every implementation plan must declare subagent assignments with tier, files, and rationale, plus any human tasks requiring approval. - Role-based model tiers: Classifies work as deep, standard, mechanical, or inherit, resolved to actual models at spawn time rather than a stale written roster. - Strict safety rules: Enforces file-level exclusivity, prohibits subagents from building, testing, linting, or committing, and protects uncommitted changes from being overwritten. - Use Case: When writing an implementation plan that spans 8 files across 3 components, use this Skill to decompose the work into non-overlapping subagent assignments, pick the right tier per task, and sequence build steps at regeneration boundaries. ## Quick Start Ask the agent to read the subagent guidelines and write the Subagent Use section for your implementation plan before spawning any subagents.
