Deep Work Plan

Deep Work Plan addresses the tendency of AI coding agents to lose focus during extended work such as migrations or large refactors. By installing a reusable skill and generating repository-specific files like AGENTS.md along with categorized documentation, it establishes explicit acceptance criteria and resumable state that any compatible agent can follow. The onboarding process examines the codebase to infer languages, frameworks, and validation commands rather than relying on generic templates. This produces an adapted harness including a cross-agent directory and verification tools that confirm conformance to the methodology. Repositories gain the ability to evolve their own skills over time while maintaining a single source of truth for plans and execution steps. The approach supports both individual projects and those coordinating across multiple repositories.