mash-multi-agent-software-harness или copilot-goal-skill

MASH (Multi-Agent Software Harness) is a structured development framework for Claude Code that manages the full software lifecycle — from idea to tested, committed code. It orchestrates four specialized AI personas (Init, Plan, Dev, QA) that work in sequence with strict boundaries: planning personas interact with you to capture intent, while implementation personas run as isolated sub-agents that write code and tests autonomously. Features are tracked through a clear lifecycle (CREATED → DEV_READY → WIP → DONE), with automatic retry logic (up to 3 attempts) and failure analysis fed back into subsequent attempts. All specs live in .mash/plan/ as the single source of truth — immutable during development. Key features: - Interactive planning that asks clarifying questions before writing specs - Isolated dev and QA sub-agents with strict read/write boundaries - Git workflow options: per-feature worktrees or current branch - Auto-commit after QA pass, or manual git control - Feature retry with failure context prop

Goal-driven task orchestration with independent verification. Interviews the user to define a clear goal, then loops between a Builder subagent (does the work) and an Inspector subagent (judges the result with fresh context). The Inspector never trusts the Builder. Output is auditable in git commits from each subagent actions. Use when the user says "achieve this goal", "make this work", "implement until done", or wants verified autonomous task completion with independent quality review.