What problem does it solve? Project requests often arrive ambiguous, span multiple disciplines, or lack explicit approval boundaries, causing AI agents to start implementation prematurely or ask redundant questions. This Skill converts raw requests into validated work contracts by separating repository facts from user decisions, enforcing approval gates before any build work, and tracking execution through structured receipts. ## Core Features & Use Cases - Automatic routing and mode selection: Detects request intent and current stage to select the minimal Work Mode, Skills, and Skill Modes without requiring users to declare them. - Approval-gated contracts: Produces a user-visible implementation brief covering intent, scope, deliverables, and validation before any change, with approval reference reuse for continuation work. - Decomposition and sequencing: Splits approved contracts into verifiable outcome units with dependency types, parallel batches, gates, and rollback plans. - Continuous work execution: Supports approved same-contract continuation with blocker taxonomy, recovery ladders, and a global progress queue. - Use Case: A user asks to add a new game feature. The Skill audits current repository state, runs a reuse-first preflight, presents an approval brief, converts the confirmed request into an executable contract, decomposes it into gated steps, and reports actual evidence at completion. ## Quick Start Ask the assistant to intake a new project request such as "route this feature request and produce an approved work contract before implementing anything."
