What problem does it solve? When delegating work to a cold-start AI session, agents often drift outside the assigned scope, guess past ambiguities, or produce artifacts in inconsistent layouts that supervisors cannot verify or archive. This Skill enforces a strict contract-based execution model so delegated tasks stay inside their brief and return verifiable results. ## Core Features & Use Cases - Brief-driven execution: Treats the supervisor's brief (Goal, Context, Tasks, Artifacts, Boundaries, Done criterion) as a hard wall, blocking scope creep and side-quests. - Standardized output shapes: Enforces exactly two output forms — a single output.md at node root, or an output/ folder with a canonical summary.md index required for archival. - Structured escalation: Defines a BLOCKED protocol that writes concrete questions to work/questions.md instead of guessing past ambiguity. - Use Case: A supervisor delegates a documentation task to a fresh AI session; the executor reads the saga plan chain, produces only the specified artifacts, classifies node-level versus saga-level outputs in summary.md, and returns a short honest status report. ## Quick Start Give the agent a self-contained brief with goal, tasks, artifacts, boundaries, and a done criterion, then ask it to execute the task as a tandem executor and return a status report.
