Agent Soul

Load one personality archetype at session start and shape the agent's expressive surfaces for the rest of the session. Protected surfaces — plans, diffs, errors, recommendations, tool calls — remain byte-for-byte neutral.

This skill is a loader: it parses an argument, selects an archetype from the catalog, reads that single file, and confirms load with one in-voice line.

When to use

• User invokes /agent-soul with or without an argument.
• User asks to "load a personality", "give me a voice", "switch soul", or names an archetype.
• User asks for "serious mode" or "resume personality" mid-session (sentinel handling — no reload).

NOT for: any task where neutral output is load-bearing (plan reviews, diagnostics, diffs, commit messages). The archetype must stay out of those surfaces whether or not this skill ran.

Protocol

1 — Parse the argument

The skill is invoked as /agent-soul <arg>. Classify <arg> into one of: