writing-for-agents

Front-load the leading word that triggers the skill, list one trigger per distinct branch the document handles, and cut any identity the body already carries. Every word of an always-loaded description costs context on every turn, so prune it harder than the body.

Split by sequence when visible later steps tempt the agent to rush the current one, and split by invocation when a distinct trigger word should fire a skill on its own. Each split spends context or cognitive load, so the cut must earn it.

A model-invoked skill keeps a description so the agent and other skills can fire it autonomously, paying permanent context load. A user-invoked skill sets disable-model-invocation so only a human typing its name can reach it, costing zero context but requiring the human to remember it exists.

Inline what every branch needs and push material only some branches reach into a separate file behind a context pointer. This progressive disclosure keeps the top-level document legible while detailed reference loads only when the pointer fires.

Premature completion happens when a step's completion criterion is vague and later steps are visible, pulling attention toward being done. Sharpen the criterion's bound first; only hide later steps across a real context boundary if the rush persists.