using-agent-expressionlisted

# Using Agent Expression Before writing: 1. **Name the reader.** Identify who will encounter the result and what they need from it. 2. **Set the done condition.** Write down what will exist when the task is complete and what evidence — a test, a command, a read-back — will prove it. 3. **Write the result.** Keep the work of producing it in the working channel. The result should stand on its own for its reader. Load one focused skill when the task needs more guidance: - `writing-for-the-reader` for any durable text. - `deciding-before-you-write` when scope, evidence, or the next action is uncertain. - A `writing-*` application skill when its name matches the work in front of you. Use the smallest set that covers the task. The skills cooperate with the reader's actual context and the project's own conventions. Routing stays in working state. Apply the selected conduct in the result instead of narrating skill names or the routing decision.