What problem does it solve? Complex engineering initiatives handed to an AI agent as a rough brief often fail because the prompt lacks verified facts, authority boundaries, completion criteria, and verification steps. This Skill turns informal briefs, links, and expected outcomes into self-contained master prompts that another capable agent can execute for hours without inventing product facts, permissions, or a definition of done. ## Core Features & Use Cases - Brief normalization and evidence labeling: Converts rough briefs into structured objectives, constraints, and terminal conditions, labeling every material fact as VERIFIED, INFERRED, ASSUMED, UNKNOWN, or BLOCKED. - Technique selection by failure mode: Chooses among RCTF, ReAct, Tree of Thoughts, Chain of Verification, self-consistency, RAG, and spec-driven development based on the failure each closes and its cost in extra passes. - Unattended run contract: Builds prompts with machine-checkable completion conditions, write boundaries, unknown-handling rules, and failure definitions so long autonomous runs survive session limits. - Use Case: You need to hand a multi-repository migration to another agent. Provide the brief and repository links, and receive a copy-paste-ready master prompt with alternatives analysis, safety boundaries, spec-driven artifacts, rollout phases, and a completion contract. ## Quick Start Ask the agent to use the master-prompt-builder skill to turn your initiative brief into an evidence-backed implementation master prompt.
