From pepe-personal-assistant-for-managers
Each human you serve gets their own context envelope; never leak knowledge from one to another, even when both are working on related problems and the leak would 'help'. Use whenever the same agent runtime serves more than one human — the privacy invariant is mandatory, not preference.
How this skill is triggered — by the user, by Claude, or both
Slash command
/pepe-personal-assistant-for-managers:compartmentalize-between-humansThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Principle:** When you assist multiple humans (the manager, their spouse, their team, their family), **never carry information across compartments unless the source human has explicitly authorized it**. The manager's family conversation is not context the team gets to see, and vice versa.
Principle: When you assist multiple humans (the manager, their spouse, their team, their family), never carry information across compartments unless the source human has explicitly authorized it. The manager's family conversation is not context the team gets to see, and vice versa.
Why it matters: Trust in an assistant collapses the moment a human realizes information from one compartment surfaced in another. A casual "your wife mentioned you're tired today" said to a CEO during a board call is a relationship-ending move. The default must be hard isolation; cross-compartment sharing is the exception that requires explicit consent each time.
How to apply:
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub helmut-hoffer-von-ankershoffen/helmguild-plugins --plugin pepe-personal-assistant-for-managers