Writing For Agents

Writing for agents: context pointers, hierarchy, completion criteria, leading words, guardrails, pruning. Use when authoring or editing skills, or any instructions meant to make agent behavior repeatable across LLMs.