What problem does it solve? Distributing a shared AI-agent scaffold (rules, hooks, skills, symlinks for Claude Code, Codex, and Copilot) into other repositories is error-prone: manual copying risks clobbering existing files, breaking symlink layouts, or overwriting a repo's own rule directories. This Skill turns that into a guided, safe UPSERT workflow. ## Core Features & Use Cases - Interactive multi-phase sync: asks which AI tools to configure (Claude Code, Codex, Copilot), whether to include .NET solutioning, and how to handle file conflicts (global overwrite vs. selective). - Rules-layout pre-flight: detects when .agents/rules is a real directory instead of a symlink to .github/instructions and stops before the sync would destroy the repo's own rules. - Deterministic sync scripts: scripts/sync.sh and scripts/sync.ps1 execute the copy/symlink work idempotently, with --template-url/--template-ref support for pinned remote clones and a --rules-only mode that distributes just the rule system as a versioned package. - Use Case: You maintain a template repo of agentic coding standards and want to roll it out to ten existing projects without deleting any project-specific files — run the skill, answer three questions, and the scripts merge the scaffold additively. ## Quick Start Ask the agent to sync the smooth-ai-stockanalysis agentic scaffold into your current repository, choosing which AI tools to configure and how to handle existing file conflicts.
