oliveryasuna/ai ? reverse-engineered prompt

Build me a small personal toolkit repo for Claude Code that I can use as my source of truth for custom skills, slash commands, subagents, and a few internal docs.

I want a simple sync script that can take the folders in this repo and link them into my ~/.claude setup, or into another location if I set an environment variable. It should preview changes first, handle conflicts safely, and let me force replace things when I really want to.

Also include a few example skills for common workflows, like pausing for deliberate decisions, giving both AI time and human time estimates, and doing a guided codebase exploration at the start of a session. Add the supporting docs so it is easy to understand how to use the skills, commands, and agents, and include a short guide for running Claude Code in a sandbox with shared config. If you need current Claude Code details, look them up online first.