What problem does it solve? Documents written for AI agents—skills, AGENTS.md, CLAUDE.md, sub-agent prompts—often cause unpredictable agent behavior because of weak trigger wording, bloated bodies, vague completion criteria, and stale content. This Skill provides a systematic reference for writing and reviewing such documents so the agent follows the same process every run. ## Core Features & Use Cases - Context pointer design: Write descriptions and pointer lines with front-loaded leading words and one trigger per branch, balancing context load against cognitive load. - Information hierarchy and progressive disclosure: Decide what stays inline, what moves behind a pointer, and when to split a document by sequence or invocation. - Completion criteria and review modes: Define clear, demanding done-conditions per step, and audit documents against seven failure modes (premature completion, duplication, sediment, sprawl, no-op, negation, ambiguity). - Static audit tooling: Run the bundled audit.py to check skills against measurable rules, detect CLI documentation drift, and find dangling cross-references. - Use Case: When writing a new skill or reviewing an existing CLAUDE.md, invoke this Skill to sharpen the description's trigger wording, push reference material behind pointers, and verify each step ends in a checkable criterion. ## Quick Start Ask the agent to review your SKILL.md or CLAUDE.md using the writing-for-agents guidelines and report each of the seven failure modes as present, absent, or not applicable.
