What problem does it solve? When an orchestrating AI (the Architect) delegates implementation work to a background coding agent (the Executor), the Executor often asks questions, drifts out of scope, or reports success without verifying anything. This Skill gives the Executor a strict behavioral contract so delegated tasks complete autonomously and report results in a machine-parseable format. ## Core Features & Use Cases - Zero-Interactivity Rules: Instructs the Executor to never ask the user questions and to make reasonable technical decisions independently based on the Architect's instructions. - Structured Final Report: Mandates a Final Report with Outcome, Files changed, Verification, and Issues sections, formatted precisely so the Architect can parse it programmatically. - Scope and Blast-Radius Discipline: Enforces staying within focus_files and task_instructions, investigating unfamiliar changes before overwriting, and documenting forced deviations under Issues. - Use Case: An Architect AI delegates a bug fix to a headless coding agent over MCP or a CLI wrapper; the Executor reads the focus files, implements only the specified change, runs the actual test suite, and returns a parseable Final Report marking the outcome as completed, partial, or failed. ## Quick Start Apply the headless-executor-contract when running as a background Executor agent so you follow its zero-interactivity, scope discipline, and Final Report rules for the delegated task.
