Optimizing Skill Descriptions

Operational guide for auditing and rewriting SKILL.md description fields so they trigger the way mizchi wants them to.

The upstream reference is agentskills.io/skill-creation/optimizing-descriptions. This skill keeps the framework intact and adds the mizchi-specific two-track policy on top.

Why this exists

Agents see name + description at startup (progressive disclosure) and decide from that alone whether to load the full SKILL.md. The description is the entire triggering contract:

• Under-specified → skill never fires when it should.
• Over-broad → skill fires when it shouldn't, wasting context.

Both failures are silent — the skill simply isn't doing its job. This audit makes them visible.

When to invoke

Meta-skill, explicit invocation only. Triggers: