What problem does it solve? AI agents start every session with zero knowledge of a project, forcing them to re-derive the codebase from scratch each time. This Skill builds a persistent context/ directory and AGENTS.md hierarchy so agents load only the slice of project knowledge a task needs. ## Core Features & Use Cases - Project Scanning and Classification: Reads manifests, READMEs, and config files to classify the project type (SPA, API, monorepo, serverless, CLI) before writing anything. - PROSE-Based Tree Design: Applies five architectural constraints (Progressive Disclosure, Reduced Scope, Orchestrated Composition, Safety Boundaries, Explicit Hierarchy) to keep every context file under ~8KB and scoped to one task type. - Context Map and Approval Gates: Wires a root AGENTS.md with a Context Map routing task types to the right files, plus explicit approval gates for risky operations like migrations and deploys. - Use Case: After inheriting an unfamiliar monorepo, run this Skill to generate context/project-overview.md, context/code-standards.md, an architecture decision log, and nested AGENTS.md files so any future agent session onboards in minutes. ## Quick Start Ask the agent to set up context files for this project so future sessions understand the codebase.
