What problem does it solve? Coordinating multiple AI providers and agents across a complete software project—from empty folder to production and retirement—lacks structure, approval gates, and verifiable evidence. This Skill provides a portable control plane that routes requests to the correct lifecycle phase, enforces human approval at consequential gates, and persists recoverable project state. ## Core Features & Use Cases - Automatic phase routing: Classifies natural-language requests and selects the earliest missing lifecycle phase (governance, discovery, specification, architecture, planning, implementation, verification, release, operations, retirement) using route_phase.py and repository state inspection. - Provider-aware delivery: Coordinates local agents, Codex, Antigravity, Google Jules, NotebookLM, and Stitch with explicit boundaries, registered transports, and independent verification of all work. - Approval and audit gates: Requires machine-readable execution contracts, signed human approvals, and durable artifacts before dispatch, release, or destructive actions. - Use Case: A user says "build a study planner app from this empty folder." The Skill runs governance and setup, guides discovery and specification, plans an approved task DAG, dispatches bounded implementation, and gates release behind independent assurance. ## Quick Start Use $agentic-engineering to audit and adopt this repository, then route my next product request through the full lifecycle.
