What problem does it solve? Dispatched agent sessions cannot ask follow-up questions, so an incomplete or ambiguous prompt causes them to fail, leak their full transcript upward, or act outside their intended scope. This Skill guides the authoring of prompts that are complete, literal instruction sets a session can execute autonomously. ## Core Features & Use Cases - Self-contained prompt construction: Ensures every rule, input, job definition, and output shape is stated directly in the rendered prompt. - Explicit substitution slots: Requires naming every placeholder token and what it resolves to, so no rendered prompt ships with unfilled variables. - Return-value contracts: Defines the exact shape of what the session sends back, typically a single status line, keeping working context inside the session. - Use Case: When building a pipeline that dispatches parallel implementation agents, use this Skill to author each agent's prompt with explicit file paths to read, hard boundaries on what it may write or execute, and a compact return contract. ## Quick Start Author a dispatched-session prompt for an implementation agent that reads its task files itself, states its write boundaries, and returns only a single status line.
